Comparison of pollutants’ load in eroded particles with those of different sewer sediments
In order to awhich type of deposits contributes to wet weather flow pollution, a comparison between the different sediments characteristics and those of wet weather flow was carried by previous researches.
In Brussels’ main trunk, Verbanck (1992) noticed that the particles transported in suspension during rainfall events are of different nature than those of type A deposits.
For 4 rainfall events on Zurich catchment, Krejci et al. (1987) estimated that 20% of suspended load and 23% of COD load is due to the erosion of biofilms whereas the other types of sediments contribute by 39% of the total TSS load and 35% of the COD load.
The fluid sediment identified in Ashley et al. (1994) in the Dundee sewer was found to be highly concentrated with COD (87.5-193 g/l) and was believed to contribute significantly to any foul flush. Ristenpart et al. (1995) realized a clear interaction between the sewage flow and the fluid sediment layer in Hildsheim interceptor at the beginning of the rainfall event where the COD concentrations in the wet weather flow were similar to those in the fluid sediment layer just after the rising limb of the rainfall event. This layer was considered to be responsible of the marked increase in the pollutants’ load transported with the sewage, the famous phenomenon of “first foul flush”. Verbanck (1995) also highlighted the role of his dense undercurrent in the “foul flush” phenomenon in combined sewers.
Ahyerre (1999) carried out an extensive analysis of the mass and characteristics of sewer deposits and compared them to the eroded particles on both sewer scale and pipe scale. He noticed that the only sewer sediment whose mass and characteristics are of the same order of magnitude as those of the eroded particles in wet weather flow is the organic layer. Type A was found to be more mineral than the erode particles and the biofilms presented much higher metallic content than those in the eroded particles (Table 1-11).
There is a clear consensus on the considerable pollution potential of this organic near bed sediment and its major role in the wet weather erosion. However, various results were obtained concerning the nature of this near bed sediment. This might be attributed to the different sampling method implemented and different characteristics of the studied sites. Table 1-12 below summarizes the different field studies found in the literature with the different techniques used and the nature and characteristics of this sediment layer observed.
It is remarkable that the type of the organic near bed sediment is tightly linked to the sampling technique used to identify and characterize it. On sites where bed traps were used (Dundee 1, 2, and 3), this sediment layer was supposed to be composed of organic particles moving near the bed that were called “Near Bed Solids”. On other sites where a multidepth sampler was used (Hildsheim, Brussels), a fluid sediment is observed. Even on the Marais catchment, when Ahyerre (1999) determined the vertical profile of SS concentration using the multidepth sampler, the near bed concentrated area was supposed to be the fluid sediment observed on other sites. However, when Ahyerre (1999) then installed his observational box at the same points where this fluid sediment layer was identified, he found that this fluid sediment is only a result of the aspiration of particles from the immobile organic layer at the bottom and not from the overlying flow. So, a new investigation of the other sites was performed by Oms (2003) using the endoscope device in Dundee, Hildesheim, Marseille, and Yorshire. Particles of organic nature floating on the surface of the coarse deposit layer were seen by the endoscope in Dundee and Hildesheim which correspond to the NBS and organic near bed fluid sediment detected before by Arthur (1996) and Ristenpart et al. (1995) respectively. The agreement of field observations between different sampling methods lowers the probability of the sampling technique being involved in the sediment deposit identification.
So, other factors related to the site characteristics were suspected. The collectors in which the organic layer was found can be classified in an upstream level compared to the collectors sampled on the other sites where fluid sediment or NBS were detected. The similar chemical composition of the different kinds identified for this near bed sediment suggests that all are of the same origin and made up of the same material but take different forms according to the hydraulic conditions dominating their occurrence position.
On the very upstream trunks of the sewer that are very close to the input sources and have very low flow velocity, these organic materials that are not still decomposed will settle down and will be specially trapped inside the coarse sediment cavities forming the immobile organic layer. The organic particles that fail to settle at these parts will be transported further downstream and decompose along its way forming near bed solids that due to the high flow velocity in downstream zones will remain in suspension or move by silting, sliding, and rolling. These particles could sometimes occur in a continuous layer moving near the bed or separate particles moving as a bed load. The distinction between these two forms is complicated. The deposits observed in Dundee by Arthur et al. (1996) and supposed to be near bed solids were also called organic near bed fluid sediment in the same article to avoid any confusion with the traditional definition of the bed-load given by Ackers et al. (1994) cited in Arthur et al. (1996) i.e. “The part of the sediment load which travels by rolling, or sliding, along the sewer invert, or deposited bed, or by saltating”. The same observations made in Dundee of sewer sediments were also studied in Ashley et al. (1994) who called it fluid sediment since as stated therein: “Despite the evidence to support the use of term “bed load” to describe the sediment collected in the trap studies in Dundee, there is some disagreement as to whether material moving close to the bed of a sewer is strictly bed-load…layering of this material has been detected giving problems for echo-sounding.”. So, the debate about the adequate descriptive term of the prevailing highly organic material near the bed in combined sewer is still raging.
However between the organic layer and the NBS/fluid sediment, the current knowledge does not permit to decide whether both exist simultaneously (immobile layer with an upper layer composed of particles detached from this immobile layer but are still heavy to be carried in suspension and thus move near the bed) as seen by Hemmerle (2014) or each form exist individually depending on the hydraulic conditions as explained before.
On other sewers with steeper invert slope as in Ecully (Lyon) (average bed slope = 2.7%), the sewage flows at higher velocity leaving no chance for these organic materials to develop in its trunks in neither an immobile organic layer nor a fluid sediment or near bed solids.
Dynamics of sewer sediments
After observing and characterizing the different deposit types formed inside combined sewers, researchers were interested in studying the dynamics of these sediments to understand the processes occurring during dry and wet weather conditions.
Accumulation of sewer grits
Although their characteristics were different from the wet weather effluents and thus are unlikely to contribute to discharges pollution, sewer grits have made the subject of extensive field studies as they were believed to influence the morphodynamic and hydraulic properties of sewer channels.
A two year survey of in-sewer sediment in collector 13 in Marseille (Laplace, 1991) provided useful elements for the comprehension of the dynamic of in-sewer coarse sediments during dry and wet weather periods. The majority of the studied events generated an increase in the deposits volume in collector 13. The author explained this phenomenon by the low energy of water flow and thus the transport capacity in this part of the sewer favouring deposition of the particles eroded during the same rainfall event from the upstream parts of the sewer system. It was demonstrated through the measurements of the evolution of deposits profile with time that sewer flow tends to distribute the deposit particles along the collector according to a decreasing depth from the upstream to the downstream thus steepening progressively the invert slope of the collector (from 0.001 m/m to 0.0035 m/m). Rainfall events disrupt the regularity of this profile by abrupt lateral inflows through junctions and gully pots provoking local erosion at these points. Then, the successive dry weather flow corrects these irregularities and smoothes the grade line by filling up these depressions. This phenomenon of progressive steepening of invert bed towards an equilibrium slope was explained by the increase of the average flow velocity due to the increase in bed slope and thus to an increase of transport capacity that will consequently prevent further accumulation in the parts previously favourable for deposition.
In their long term monitoring of sewer sediments in the Lacassange sewer (from 2000 to 2004), Bertrand Krajewski et al. (2006) found an asymptotic increase of both sediment mass and longitudinal slope (Figure 1-1) with time corroborating the results of Laplace (1991). This tendency was explained by the associated increase in flow energy capable of transporting more particles or by the balance occurring between erosion and sedimentation processes.
Erosion of organic near bed materials
From a theoretical point of view, the erosion of a deposited particle is mainly due to the transfer of mean and turbulent kinetic energy of water flow to particles on the flow bottom (Mehta et al., 1989; Hérouin, 1998) by means of lift and drag forces. Lift forces are caused by pressure and viscous skin friction and is proportional to the square of the flow speed, while drag forces are caused by the velocity gradient in the flow (shear effect) and by the spinning motion of the particle (Magnus effect) which is considered to be in viscous flow and turbulent flow proportional to the flow speed and to its vertical gradient (Van Rijn, 1984). In the in-sewer sediment case, the cohesion property of these organic deposits induces an opposing force to these forces, and thus particles motion does not start before the destabilizing forces exceed the sediments’ cohesive resistance. Current knowledge of the mechanism of these organic deposits including the erosion process and its rate and the cohesive resistance is limited to the few experimental studies conducted in sewer systems either in real conditions (dry and wet) or in artificial context (flushing experiments). The scarcity of such studies is related to the onerous procedures necessary in field investigations when working in the hostile hardly controllable and heterogeneous environment of sewer systems.
Investigations in real sewer flow conditions (dry/wet)
Measurements on several combined sewer systems (Brussels main trunk (Verbanck, 1990), Marais (Gromaire, 1998), Quais and Clichy (Lacour, 2009), Ecully (Métadier, 2011)) show a daily cycle of both flow rate and TSS concentration (Figure 1-2). When the cycle of the flow rate seems logical due to the fluctuations of water consumption during the day, the cycle of the TSS concentration is more surprising. This evolution of TSS concentration was exhaustively investigated in (Gromaire, 1998) by analysing the data acquired in all the dry weather measurement campaigns. It was found that the increase of TSS concentration is accompanied by an increase of the discharge flow due to the increase of the transport capacity of the flow regime that triggers particles’ erosion more than being a result of dilution by clean water. This was concluded from several elements observed during the measurements at the peak flow moment: increase in particulate proportion with wastewater flow, decrease in the organic content, decrease in the biodegradability of the effluents, and sharp increase in the ratio TSS/CODd along with a break in the relation between TSS transport rate and wastewater flow rate.
The temporal monitoring of the “dense layer” during dry weather periods as a function of hydrodynamic conditions of sewage flow, Hemmerle (2014) noticed that the build-up of this layer starts at the moment following the morning peak flow after being eroded by this latter, where in more general expression, a flow speed > 0.13 m/s corresponding to a shear stress of 0.055 N/m2 is sufficient to initiate the suspension of this layer.
Gromaire (1998) has studied the variation of the contribution of the sewer deposits inside the storm event and showed that this contribution occurs along the whole duration of the event but with a highly variable proportion from one moment to another. It’s maximum at the moment of peak flow rate and in the following 30 minutes due to the erosion process. In events having several flow peaks, a net decrease in the TSS production is observed between the consecutive peaks demonstrating a progressive decrease of the sediment stock available for erosion, possibly related to an erosional strength of the contributing deposits increasing with depth. The organic content of the particles issued from the deposits erosion was high throughout the whole rainfall event even at the moment of the flow peak which signifies that even when the flow energy increases no (or little) erosion of the mineral deposits takes place. The maximum concentration of volatile matter was generally observed at the beginning of the rainfall event whereas the minimum values were obtained at the end of the rainfall event.
Another study conducted by (Ristenpart et al., 1995) based on the comparison of the vertical SS concentration profiles between dry weather flow and wet weather flow indicated that the interaction of the organic near bed fluid and the overlying flow is the phenomenon responsible for organic particles production during wet weather. For an extensively monitored storm event on Hildesheim, the well-marked SS concentration gradient at the beginning of the rainfall event decreased significantly during the rise of the water level to become fairly homogeneous where the SS content was doubled in the sewage flow and halved in the fluid sediment indicating a suspension of the fluid sediment into the water column. A few minutes later just at the moment of the flow velocity peak, a short time scale increase in the SS concentration was noticed due to rapid erosion not only of the fluid sediment layer but also of the sediment bed itself as the deposits bed decreased.
Investigations in controlled flushing experiments
A flushing experiment was carried out on the low-sloped Brussels main trunk by Verbanck (1995) as an alternative approach to the heavy experimental campaign made during real rainfall event on the sewage flow and sediments layer to compare their properties and identify their interaction mechanism. An artificial change in the hydraulic conditions (flow rate and shear stress) due to the water release from the vane of a cleaning trolley displayed a marked increase in the SS concentration determined at mid-depth of the flow section. The shear stress induced by this experiment was between 1.1-1.2 N/m2 at the mid-depth giving a local shear value of 1 N/m2 at the level of the dense undercurrent layer. Such low value affirms that the eroded layer is looser than the coarse granular deposits that need high shear stresses to be scoured.
Table of contents :
Part I State of the art & dataset presentation
Chapter 1 State of the art
1. Pollution of Urban Wet Weather Discharges (UWWD)
2. Origin of pollution in Urban Wet Weather Discharges
3. Sewer deposits
3.1. Composition and Characteristics
3.1.1. Sewer grits
3.1.3. Organic near bed sediment
220.127.116.11. Type C deposit
18.104.22.168. Near Bed Solids
22.214.171.124. Organic layer
126.96.36.199. Fluid sediment
3.1.4. Comparison of pollutants’ load in eroded particles with those of different sewer sediments
3.2. Dynamics of sewer sediments
3.2.1. Accumulation of sewer grits
3.2.2. Erosion of organic near bed materials
188.8.131.52. Investigations in real sewer flow conditions (dry/wet)
184.108.40.206. Investigations in controlled flushing experiments
4. Modelling sewer processes
4.1. Modelling in-sewer sediment erosion
4.2. Modelling suspended solid transfer in sewer system
5. Modelling hydraulics in sewer systems
System Using an Adapted Hydrodynamic Model
5.1. Simple hydraulic models
5.2. Complex hydraulic models
5.3. Numerical schemes of Shallow Water Equations
5.3.1. Flux term discretization
5.3.2. Source term discretization
Chapter 2 Presentation of experimental and modelling investigations of the Marais site
1. Choice of the study site
2. Site description
2.1. Surface description
2.2. Sewer system description
3. Database presentation
3.1. Precipitation data and rainfall events characteristics
3.2. Dry weather flow data
3.2.1. Discharge flow rate
3.2.2. TSS concentration
3.3. Wet weather flow data
3.3.1. Runoff flow
3.3.2. Sewer outflow
3.4. In-sewer deposits characteristics
3.4.1. Type of deposits
3.4.2. Localization and topography of sewer deposits
3.4.3. Pollutants’ content
220.127.116.11. Experimental results of the dynamics of the organic layer
18.104.22.168.1. Controlled flushing experiments
22.214.171.124.2. Observational system
126.96.36.199. Results of modelling the dynamics of the organic layer
Part II The interest of a distributed hydrodynamic model specifically adapted to collectors of highly varying bed slope and cross-sections
Chapter 3 A Simple Finite Volume Method for 1D Naturally Balanced Shallow Water Equations
Chapter 4 Sensitivity of Hydrodynamic Models and Solid Transport to the Description of Silted Collectors
Part III In-sewer suspended solids sources and solid transport modelling
Chapter 5 Do storm events samples bias the comparison between sewers deposits contribution?
Chapter 6 Development and Benchmarking of Solid Transport Models with Different Scenarios for In-sewer Sources of Suspended Solids
General Conclusion and Perspectives