Case study of H₂S release and transport in a trunk sewer with drops

The City of Edmonton, Alberta, Canada, like many other cities worldwide, is facing sewer odor nuisance and pipe corrosion issues. The main reason for the odor complaints and pipe corrosion problems is hydrogen sulfide (H₂S) generated in sewer systems (Hvitved-Jacobsen et al. 2013). When dissolved oxygen and nitrates are depleted as bacteria metabolize organic material in sewage, sulfate (⁠⁠) is utilized and reduced to sulfide (S²⁻) by sulfate-reducing bacteria (SRB). The SRB mainly belong to eight genera: Desulfovibrio, Desulfotomaculumn, Desulfomonas, Desulfobulbus, Desulfobacter, Desulfococuus, Desulfosarcina and Desulfobacterium (Visser 1995). The dissolved H₂S in sewage can be released into the sewer headspace above the sewage and transported by the air flow, causing sewer odor complaints when it emits to the community. Meanwhile, the H₂S in the sewer headspace can be absorbed by the moist concrete surface in the pipe and oxidized to sulfuric acid (H₂SO₄) by sulfur oxidizing bacteria (SOB) with the presence of oxygen and cause corrosion problems (Gomez-Alvarez et al. 2012). Thiobacillus were found to be the key community members of SOB (Hernandez et al. 2002). Besides, Thiothrix, Thiomonas intermedia, Halothiobacillus neapolitanus, Acidiphilium acidophilum, and Acidithiobacillus thiooxidans were also found to be SOB responsible for concrete corrosion (Okabe et al. 2007).

What makes Edmonton's sewer system special is the large number of drop structures (over 800) due to the rather deep trunks (up to 50 m below the ground) in the sewer network. Sewer structures such as junctions, manholes, bends, weirs, and drops may give rise to increased turbulence compared with the hydraulic conditions that exist under normal flow conditions in a sewer pipe. In particular, at sewer falls and drops, the mass transfer between water and air is significantly increased, caused by phenomena such as splashing droplets and entrainment of air in the water phase (Ma et al. 2016; Qian et al. 2017). Such structures can therefore promote H₂S stripping and oxygen reaeration (Beceiro et al. 2017; Jung et al. 2017; Guo et al. 2018). Matias et al. (2014) evaluated the influence of free-fall drops on the release of H₂S gas in a laboratory experiment and reported that the maximum H₂S concentration in the air phase can reach 500 ppm with 0.9 m drop and 10 mg/L sulfide in the liquid phase. Then further study was carried out to establish the empirical relationships between the mass transfer of oxygen and the physical parameters of drop structures (Matias et al. 2017). But these results are likely specific to particular structures and flow conditions, and their applicability in real sewer systems needs to be tested.

The emission of H₂S to the ground is directly related to the dynamics and transport of H₂S and air movement in the headspace along the system, in particular the pressurization due to the drop structures (Wang et al. 2012; Qian et al. 2018). The air movement could transport the sewer air from one location in the network to another. Once the H₂S gas is released at certain drop structures in real sewer systems, it can move with the sewer air to other locations and lead to more locations of odor complaints. It is essential to conduct the field work in a real sewer network to identify not only the generation of H₂S in the sewage, but also the stripping of the H₂S gas from sewage and its subsequent transport in the sewer network.

The Bonnie Doon area in Edmonton, Alberta, where the odor problem is prevalent, was chosen as the study area for field work. The study focuses on the sulfide emission and transport along the trunk together with the laterals. The objectives of this study are firstly to investigate the generation of sulfide in the water phase and transport of H₂S in the air phase along the trunk and relevant laterals in this area, then to evaluate the effect of drop structures on reaeration and H₂S release into the air phase in the trunk and finally, identify the functioning bacteria in the biofilm for sulfide generation and H₂S gas oxidation.

The field work was conducted in the Bonnie Doon area, a neighbourhood in south-central Edmonton, Alberta, Canada. The sewer system in the area is a combined system that carries both sewage and stormwater runoff. This study only focuses on the dry weather case. The main trunk and corresponding laterals that are connected to the trunk is shown in Figure 1(a) and 1(b). The main trunk is connected to 10 manholes in the study area, which are numbered as T1–T10, with an average flow rate from 1.257 m³/s at T1 to 1.521 m³/s at T10. There are 14 laterals (L1-L14) contributing to the trunk. Among them, L8 and L12 are major laterals with a flow rate of 0.240 m³/s and 0.290 m³/s, respectively. The flow rate of the rest of laterals was relatively low (less than 0.02 m³/s) compared to that in the trunk. In the trunk, there are three drops with a drop height of 2.7 m, 5.2 m and 2.0 m, as shown in Figure 1, which are located at the manholes T3-1, T4 and T5, respectively.

Two rounds of field monitoring were implemented in the area in 2017 and 2018. Water samples were obtained from the manholes in the trunk (T1–T10) and main laterals (L8 and L12) for the first round on Nov. 29–30, 2017. The H₂S gas concentration and air pressure at the sewer headspace of the trunk (T1–T10) and laterals (L1, L3, L4, L5, L6, L8, L8-1, L9, L10, L12, L13) were monitored continuously by Odalog (App-Tek, Queensland, Australia) and SmartReader (ACR Systems, Vancouver, British Columbia, Canada) from Nov. 29th to Dec. 6th, 2017 and Dec. 13th to Dec. 20th, 2017. Odalog was directly suspended 2 meters below the manhole cover using a steel cable. For the second round, water samples obtained from manholes in the trunk (T2–T7, T9, T10) and main laterals (L8 and L12) on Sept. 27th, 2018. Besides, water samples were also taken from manholes (L2-1, L3-1, L4-1, L6, L7, L10, L11-1) before drops in the laterals on Sept. 19th, 2018. Odalog and SmartReader were installed in the trunk (T1–T7, T9–T10) and laterals (L1, L2–L14) from Sept. 18th to 25th, 2018 and Sept. 27th to Oct. 4th, 2018. All water samples were collected in planned locations. The dissolved oxygen (DO), pH and temperature of the water samples were measured on site by portable pH meter and DO meter. The samples for total and dissolved sulfide (TS and DS) analysis were preserved on site too. Dissolved and suspended sulfide were separated by precipitation after addition of aluminum chloride to the wastewater sample, which produced an aluminum hydroxide floc that trapped suspended sulfide. The samples for total sulfide determination were preserved by addition of 2M zinc acetate (2 mL/L) and 6N sodium hydroxide (2 mL/L) solution to the sampling bottles. All the field water samples were stored in an ice-box during transportation to the laboratory and were kept refrigerated till analysis was done.

The water samples were analyzed for the determination of total and dissolved sulfide (TS and DS), sulfate, total, dissolved and soluble chemical oxygen demand (COD) (TCOD, DCOD and SCOD), volatile fatty acid (VFA), nitrate (NO₃), ammonia (NH₄), total nitrogen (TN), total phosphorus (TP), total suspended solids (TSS) and volatile suspended solids (VSS). Total and dissolved sulfide were analyzed according to the methylene blue method (APHA et al. 2017). The sulfate, total, dissolved and soluble COD, nitrate, ammonia, TN and TP in wastewater samples were analyzed according to Standard Methods (APHA et al. 2017). DCOD was measured firstly by filtering the water samples through 0.45 μm membrane. All the parameters analyzed in our laboratory were measured twice (variation was less than 5%). The water samples for the determination of VFA, SS and VSS were sent to a commercial laboratory (CARO, Edmonton. Canada) for analysis according to Standard Methods (APHA et al. 2017) which were measured in triplicate. The wastewater samples from the trunk (T1–T10) and main laterals (L8 and L12) were analyzed to evaluate the general water characteristics in this area, as is shown in Table 1.

The biofilm samples were collected in the first round by scraping from the manhole wall, using a spoon, 5 m above the bottom at the manholes T1 and T2, and from sewer wall above the water level interface at T1 and L8. For sediment collection, the sediment was quickly grabbed at the bottom of the trunk at T1. All the biofilm and sediment samples were analyzed through 16S rRNA for identification of the microbial community. The total genomic DNA was extracted from the collected biofilm and sediment samples by using the MO BIO PowerSoil DNA isolation kit (QIAGEN, Toronto, CA) according to the manufacturer's instructions. Extracted DNA was stored at −80 °C until further processing. The 16S rRNA gene fragments were amplified from the extracted total DNA with a primer set (515F/806R) targeting the V4 hypervariable region of both the Bacteria and Archaea domains (Caporaso et al. 2012). The sequencing was performed on the Illumina HiSeq 2500 platform at BGI Genomics Inc., Shenzhen, China. The raw data were filtered to eliminate the adapter pollution and low quality to obtain clean reads, then paired-end reads with overlap were merged to tags. The tags were clustered to Operational Taxonomic Units (OUT) at 97% sequence similarity. Taxonomic ranks were assigned to OTU representative sequence using Ribosomal Database Project (RDP) Naive Bayesian Classifier v.2.2. Lastly, alpha diversity, beta diversity and the different species screening were analyzed based on OTU and taxonomic ranks.

• C is the concentration in the liquid phase (g/m³)

• C_S is the saturation concentration of a particular gas in the liquid phase (g/m³)

• K_La is the overall mass transfer coefficient at (h⁻¹).

• C_u is the concentration at the upstream of the sewer drop (g/m³)

• C_d is the concentration at the downstream of the sewer drop (g/m³)

• t_f is the falling time (h)

• t_w is the residence time of the tailwater (h)

• H is the drop height (m)

• g is the gravity acceleration constant 9.81 m/s²

The residence time t_w is estimated using the distance between the two drops divided by the flow velocity, which stands for the travel time of turbulent sewage between two drops. Note that the inclusion of this residence time here is to account for the additional mass transfer occurred after the impingement of the falling sewage on the bottom and the enhanced turbulence during the transport in the trunk. The determination of this residence time is somewhat uncertain.

The water samples from the trunk and major laterals were analyzed for the sulfide (TS and DS) and sulfide-related parameters and the results are presented in Table 2. The total sulfide (TS) and dissolved sulfide (DS) at T1 were 1.43 mg/L and 0.42 mg/L, respectively, in the first round, which was formed at the upstream section of the trunk. The DS concentration in the two major laterals, L8 and L12, was low in the first round (0.00 and 0.31 mg/L) and zero in the second round. Therefore, only the sulfide built up at the upstream sections of the trunk mainly contributed the sulfide in the trunk.

The evolution of sulfide along the trunk can be seen in Figure 2. Both rounds of monitoring results show a consistent trend. The sulfide (TS and DS) increased from T1 to T3, while the dissolved oxygen (DO) slightly dropped. After T3, the sulfide fell all the way until the end of the trunk (T10) except for T8 in the first round and DO increased along the trunk until T9, then decreased again from T9 to T10. Since the retention time between T7 and T8 was rather short (0.9 min), it did not allow the formation of sulfide. The abnormal rise at T8 resulted from the diurnal variation. On the other hand, H₂S in the air phase along the trunk is shown in Figure 3. H₂S was present at manholes of T3, T4, T5, T6, T7, T8 and T9. H₂S gas was detected over the day with a diurnal variation, and the highest value occurred around 12:00 AM. However, H₂S gas in the manholes of T1, T2 and T10 was zero. The air temperature was 13–15 °C in the first round and the air temperature ranged from 9 °C to 13 °C in the second round (Figure 3).

As shown in Table 2, in the first round, DS concentration rose from 0.42 mg/L to 0.83 mg/L and DO concentration decreased from 1.5 mg/L to 1.2 mg/L from T1 to T3. The retention time along the nearly 1,500 m trunk distance (T1–T3) allowed the generation of sulfide by sulfate-reducing bacteria (SRB) in the biofilm on the pipe wall and then transferred into the water phase under the low DO level. The drop of DO was because of the activity of aerobic bacteria. At the beginning of the trunk, the flow at T1 and T2 was under normal flow conditions where the flow velocity was around 1.6 m/s. The DS in the sewage was low (less than 1.0 mg/L) and not much H₂S gas can be transferred into the air phase. The transferred H₂S gas from the sewage can be completely absorbed and oxidized on the moist sewer and manhole wall in the deep trunk (around 26 m deep). There was no H₂S at manholes T1 and T2. However, it should be mentioned that H₂S was detected at 1 ppm and 1–2 ppm at T1 and T2 in the first 10 minutes when the manhole covers were opened to install the Odalog and SmartReader. More sewer air moved to the top of the uncovered manholes, which brought more H₂S gas to the top. The H₂S could not be completely removed by adsorption process. This phenomenon indicated that conditions close to equilibrium of H₂S between air and water phase rarely exist in real sewer networks because of adsorption and oxidation processes at sewer and manhole walls.

From T3 to T4, the concentration of TS and DS decreased to 1.42 mg/L and 0.70 mg/L, respectively, and DO increased to 3.2 mg/L at T4. The retention time from T3 to T4 was also too short (0.8 min) to generate sulfide. And the trunk sewer drops at T3-1 (2.7 m) and T4 (5.2 m) gave rise to increased turbulence of the flow, which accelerated mass transfer of both sulfide and oxygen between water and air phase. The release of H₂S gas and reaeration of oxygen led to the decrease in sulfide and increase in oxygen in the sewage. Similarly, the trunk sewer drop at T5 with a drop height of 2.0 m further released the sulfide and reaerated oxygen in the sewage. The TS and DS dropped to 1.20 mg/L and 0.37 mg/L, respectively, and the DO increased to 6.8 mg/L at T5. The release of H₂S by these drops resulted in the high H₂S gas concentration at T3, T4 and T5. The average H₂S gas was about 15 ppm at these locations.

Because of the high DO level in the sewage after reaeration process at these drop structures, part of sulfide was oxidized. In addition, the sulfide kept being stripped into the air along the trunk (T6–T10). Sulfide concentrations in the liquid phase were observed to decrease as the wastewater continued downstream. The TS and DS at T10 decreased to 0.82 mg/L and 0.13 mg/L respectively.

By section T9, DO decrease was observed (from 7.6 mg/L at T9 to 4.6 mg/L at T10), likely due to bacterial respiration. The released H₂S in the air phase at drops was transported downstream and dropped because the process of adsorption, and oxidation of H₂S on the moist pipe and manhole wall was dramatic. So the H₂S gas concentration kept decreasing from T5 to T10 (Figure 3). The average H₂S gas concentration was 10 ppm, 3 ppm, 5 ppm, 1.5 and 0 ppm at T6, T7, T8, T9 and T10, respectively. The H₂S at T8 was higher than that at T7 likely because the air pressure at T8 (24 Pa) was higher than that at T7 (12 Pa) (Figure S2) and more H₂S gas could be moved to the top the manhole T8.

The second round of field work gave the similar results as the first round (Figures 2 and 3). The sulfide generated from the upstream sections of the trunk was not high and part of sulfide was built up at the upstream trunk (T1–T3) under low DO concentrations. No H₂S was detected at the manholes under normal gravity pipe flow (T1–T2) since the sulfide in the sewage was low (DS less than 1 mg/L). However, high H₂S gas was detected at the drops (T3–T5) under the same low sulfide concentration in the sewage and the released H₂S could be transported downstream to make them as hotspots (T6–T9). The trunk sewer drops played an important role on H₂S release, resulting in relatively high H₂S gas concentration despite the rather low sulfide concentration in the sewage.

As shown in Table 3, the H₂S and O₂ concentration in the water phase at the upstream (C_u) and downstream (C_d) of each drop based on the field measurements of the oxygen (C_O2) and dissolved sulfide concentration in the sewage (shown in Table 2). The H₂S gas (C_H2S) in the sewage then was determined by dissociate equilibrium between dissolved sulfide and H₂S gas in the sewage. H₂S gas concentration in the sewer air was measured in the field (Figure 3) and oxygen concentration in the sewer air is assumed to be 21%. Then the saturation concentration of H₂S (C_S(H2S)) and oxygen (C_s(O2)) in the sewage were determined by Henry's law with H₂S and O₂ concentration in the sewer air. The total stripping time (t) at drops consists of falling time (t_f) and residence time (t_w). t_f is the falling time at drop. The residence time t_w accounts for the additional mass transfer that occurred after the impingement of the falling sewage on the bottom and the enhanced turbulence during the transport in the trunk, which is estimated using the distance between the two drops divided by the flow velocity. Finally, the overall oxygen and H₂S mass transfer coefficient (K_La_(O2) and K_La_(H2S)) at drops in the trunk were estimated using Equation (4), and the results are shown in Table 3.

The overall oxygen and H₂S mass transfer coefficient was around 200 h⁻¹ and 300 h⁻¹ with drop height 2.7 m at T3-1 followed by a drop height of 5.2 m at T4. The oxygen in the sewage increased by approximately 3 mg/L and half of the H₂S (0.05 mg/L) in the sewage was released into the air phase, which resulted in 15 ppm H₂S gas in the sewage headspace. The overall oxygen and H₂S mass transfer coefficient was around 70 h⁻¹ and 54 h⁻¹ with a drop height of 2.0 m at T5. Combining this additional drop effect, another 3 mg/L O₂ was added into the sewage after T5 and more H₂S was stripped into the air phase to almost reach equilibrium concentration in the sewage. In particular, the H₂S transfer coefficient at T5 during the second field work was positive, which indicated much more H₂S entered into the air phase and the H₂S in the liquid phase was lower than that in the air phase after the first two drops (T3-1 and T3), no H₂S was released into air phase at T5.

Compared to normal gravity pipe, the coefficient at the drops was relatively high. The oxygen transfer coefficient was in the range of 3–50 h⁻¹ in gravity sewers with different slopes and flow rates (Lahav et al. 2006). Yongsiri et al. (2004) demonstrated that a constant ratio of (K_La)_H2S to (K_La)_O2 was 0.86 ± 0.08 at 20 °C in gravity sewers, which was independent of the degree of mixing. So the H₂S overall mass transfer coefficient in the normal gravity sewer is also in the same range as the O₂ transfer coefficient. In addition, the overall mass transfer coefficient dramatically increased with the drop height. The number from T3-1 to T4 was much higher than that at T5. Drop height has a very significant influence on reaeration and H₂S stripping efficiencies at drop structures. Matias et al. (2017) studied the effect of the characteristic of drops on reaeration and the O₂ overall mass transfer coefficient ranged from 10 h⁻¹ to 90 h⁻¹ with different tailwater depths and drop heights. The highest number was 88.7 h⁻¹, with the lowest tailwater (0.06 m) and highest drop height (1.16 m). The drops in the field are at larger scale in terms of drop height, pipe size, flow rate and tailwater and the overall mass transfer coefficient is larger compared to that of laboratory study results.

The concentration of H₂S gas in the headspace of the laterals connected to the trunk was continuously monitored and the results are shown in Figure S1 in Supplementary Material. H₂S existed at the manholes of L1, L3, L4, L5, L7, L8-1, L9 and L10. No H₂S gas was detected at L6, L8, L11, L12-1 and L13. The formed DS along the major lateral L8 was stripped into the air phase by the drop (10.9 m) at T8-L1 to make it a location of sewer odor concern. The DS concentration in the sewage at L8 was zero after release (Table 2), so the H₂S in the air phase was also zero. Liquid samples were taken from the upstream manhole of these laterals to identify where the detected H₂S came from. The results are shown in Table S1. The DS in all the sampling manholes was close to zero under high DO concentrations (>4 mg/L). The H₂S detected in those laterals was transported to these locations from the trunk. The air pressure in the trunk and laterals were continuously monitored for two weeks and the average value at each location was calculated, as is shown in Figure S2. The air pressure in the trunk was higher than that in the neighboring laterals. The air pressure in the trunk T1–T3 was around 60 Pa and the air pressure in laterals of L1, L3 and L4 was 59 Pa, 0.4 and 46 Pa. Likely T9 had the air pressure of 0 Pa, while its nearby lateral, L10, had air pressure of −4 Pa. The air pressure difference between the trunk and the laterals could push the air into the laterals, which brought the H₂S into these laterals to make them locations of odor concern.

The microbial community of the biofilm samples at different locations in the sewer system was analyzed via Illumina MiSeq sequencing. Figure 4 presents the relative abundance of the microbial community at the genus level. Desulfomicrobium, a known SRB, was identified in the biofilm on the sewer pipe wall at T1, L8 and in the sediment at T1 with relative abundance of 0.3%, 0.97% and 0.01%. Desulfomicrobium, belongs to the family of Desulfomicrobiaceae, and most of these species use pyruvate and lactate to reduce sulfate to sulfide (Watanabe et al. 2017). Thiobacillus, known as SOB, were identified as the dominant species (relative abundance >5%) in both manhole wall samples at T1 and T2 with an abundance of 14.2% and 6.51%. Thiobacillus are reported to be responsible for both producing sulfuric acid and deteriorating concrete in sewer systems (Wei et al. 2010). In terms of other samples, Pseudomonas (33.5%) was the dominant genus in the sewer biofilm sample at T1. The predominant bacterial population in the sewer biofilm sample at L8 was comprised of Arcobacter (7.3%) and Paludibacter (5.6%) as well as Acidovorax (5.3%) and Faecalibacterium (6.6%) in the sewer biofilm sample at T1. Pseudomonas are widespread in nature (water and soil) and can produce a wide range of extracellular polymeric substances, which are involved in attachment processes and biofilm formation (van Delden 2004). Arcobacter and Faecalibacterium species are highly abundant in raw sewage as pathogens (Fisher et al. 2014). The genera Paludibacter is strictly anaerobic and chemoorganotrophic, and is able to utilize melibiose, glycogen and soluble starch as growth substrates (Gronow et al. 2011). The Acidovorax genus is affiliated to the class of Comamonadaceae, which is isolated in activated sludge as denitrifiers (Heylen et al. 2008).

Only one SRB genera (Desulfomicrobium) was identified in low relative abundance in the sewer wall biofilm and sediment samples at T1 and T2. The low proportions and diversity of SRB indicate an aerobic operational environment under low oxygen with low concentrations of H₂S in the sewage system. Most of the bacteria in the sewer manhole biofilm were unclassified and Thiobacillus represented one of the dominant microbial communities, which have been implicated in the corrosion of concrete surfaces. So Thiobacillus is the major key community member contributing to sewer pipe corrosion in this area. There are severe manhole corrosion issues in the trunk, especially at the location of T3–T5 where H₂S gas concentration was high. The rest of the genera previously recognized to comprise a dramatic portion of the biofilm/sediment microbial community in the sewer were present (Gao et al. 2016; Liang et al. 2016).

Two rounds of field work were carried out in the Bonnie Doon area, Edmonton, Alberta, Canada, to study the generation, release and transport of hydrogen sulphide in a trunk sewers with three drops. The results from both rounds were similar: the sulfide generated from the upstream sections of the trunk was not high (DS less than 0.5 mg/L) and part of the sulfide could be built up in the upstream trunk (T1–T3) under low DO levels (less than 1.5 mg/L). No H₂S was detected in the manholes under normal gravity pipe flow (T1–T3) since the sulfide in the sewage was low (DS less than 1 mg/L). However, a high H₂S gas concentration was detected at the drops (T3–T5), with high turbulence under the same low sulfide concentration in the sewage with an average H₂S of 15 ppm. The released H₂S at drops was transported downstream to make them the locations of odor concern (T6–T9) for sewer odor complaints. The average H₂S gas concentration was measured at 10 ppm, 3 ppm, 5 ppm and 1.5 ppm at T6, T7, T8 and T9, respectively. The drops in the sewer system played an important role in H₂S release and rather low sulfide could still result in high H₂S gas concentrations locally. The overall oxygen and H₂S mass transfer coefficient at drops was estimated to be much higher than that in normal gravity sewers and the overall mass transfer coefficient dramatically increased with the drop height. The overall H₂S mass transfer coefficient was around 300 h⁻¹ with a drop height of 2.7 m at T3-1 followed by a drop height of 5.2 m at T4 and 54 h⁻¹ with a drop height of 2.0 m at T5. The H₂S detected in those laterals was transported to these locations from the trunk due to the air movement in the sewer network, which made some of them hotspots. Desulfomicrobium was identified as the only SRB in the sewer wall biofilm, but not in richness; Thiobacillus was the main population contributing to sewer pipe corrosion in this area, and was predominant in biofilm on the manhole wall. According to the field results, two odor control strategies could be applied in this case, either retrofitting the three drops in the trunk to reduce the stripping effect of these drops or removing the sulfide in the sewage in the trunk right before these three drops by adding chemicals like nitrate, ferric/ferrous and hydrogen peroxide etc.

The authors would like to acknowledge the financial support from EPCOR and Natural Sciences and Engineering Research Council (NSERC) of Canada. The authors would also like to acknowledge Perry Fedun for technical assistance, Guijiao Zhang, Yangbo Tang, Yu Qian, and Yiyi Ma, for their assistance with the fieldwork.

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