Curtailing COVID-19 spread in drain pipelines : Using interfacial hydrodynamics for removing Bacterial and Viral Biofilms

Drainage systems contains biological contaminants like bacteria and viruses flowing through them. Additionally, these pipelines also have organic matter known as biofilms growing on their walls. These biofilms infact act as incubation zones for further growth of bacteria and coronaviruses. Standard water treatment routines with traditional cleaning agents are known to be not be able to clean or sterilize microbes located in the inner layers of the biofilm. A recent study has identified specialised fluids which are effective in removing biofilms but these need to be used prudently. The present study proposes to use ‘ interfacial hydrodynamics’ to ensure that the cleaner-fluid (CF) is transported effectively to the location of the biofilms at the pipe walls, and allowed to be in contact with the biofilms for a sufficient amount of time so as to ensure its effective removal. The present study has used CFD technique of Multi-fluid VOF and has demonstrated that relative superficial velocities of cleaner-fluids and sewage water can be controlled, so as to achieve flow regimes that ensure delivery of cleaner fluid to the periphery of the tube walls. Our simulations indicate that most effective cleaning can be achieved by using a cleaner-fluid with a high viscosity ( ~5000 cP)) . In such cases, a low- medium velocity (~0.05-0.3 m/s) of CF and water would ensure that the cleaner fluids are in constant contact with the pipe walls. Other suitable viscosity and velocity combinations have also been proposed. Flow parameters that can be used to monitor and cross-verify expected flow patterns on-site have also been proposed.

that the computational technique used here can be an effective tool that can be used to predict the flow regime formation for drain pipeline flows, for a wide range of fluid combinations. To ensure that real-life conditions mimic what is predicted by simulations, one needs to closely monitor various parameters of the flow; and in this paper we also provide a list of fluid dynamic parameters and simple criterions that indicate adherence/nonadherence to expected flow regimes, and can be used to monitor the real field conditions. More broadly, our results suggest that high viscosity CFs are most suited for optimal cleaning of biofilms formed on pipeline walls.
It is however noted that, high viscosity CFs require higher pumping power, and thus, in places where this is difficult to achieve, operating conditions for implementation with medium viscosity CFs have also been suggested. The method suggested is highly targeted and ensures minimal usage of cleaner-fluid.

RESULTS
Considering that a cleaner-fluid (CF) would already have the chemical properties to remove the biofilms on the pipeline walls, the challenge is to ensure 'efficient' and 'sufficient' contact between the cleaner-fluid and the pipeline walls. This is clearly a fluid dynamics problem, since one particular flow regime would maximise the contact while another might not. Achieving a particular flow regime, depends on the relative fluid properties, which in this case is sewage water and cleaner-fluid. Sewage water is assumed to have properties close to water and is an invariable factor in all the combinations investigated. So the problem now depends on the cleaner-fluid property, and the single most dominant factor is its viscosity. Experimental data of liquid-liquid flows of low viscosity and high viscosity fluids is available. However, medium viscosity ranges are less explored. Thus, flow regimes achievable for medium viscosity is unknown. In this study, following the demonstration that the simulation convincingly captures the liquid-liquid flows at for both high and low viscosity CFs, the middle viscosity range was also simulated, to ensure completeness of data.

Flow morphology
Here, we present the flow morphology obtained using simulations for low, medium, and high viscosity CF cases.
In some cases where experimental observations of the flow regimes are available, the simulation results have been compared with them. Tables 1 -3 present the flow regimes obtained for low, medium and high viscosity cleaner-fluids respectively. It was observed that the flow patterns matched expected trend for most cases except when the water velocities (sewage) were low. In low water velocity cases, we know that, quantitatively, the average void fraction across a cross section is captured reasonably well by the simulation, and it the spatial distribution that is not captured accurately. This mismatch is attributed to the inability of the interfacial force models used in the CFD simulation, which are essentially empirical models, to provide the correct force quantities, when it falls below the model's applicability range. The exercise of tuning these models for improving flow regime predictions at low water velocity cases is not carried out here since water velocity in sewage lines tend to be of the medium to high velocity ranges. It is observed from the simulations that two major parameters govern flow patterns: (a) the magnitude of superficial velocity (svel) of CF and water (low, medium, high) (b) their ratio R (R= CF svel /water svel; R >1, =1, <1). Superficial velocity is a hypothetical flow velocity calculated as if the given phase or fluid were the only one flowing or present in a given cross sectional area. This is the common metric of reporting operating conditions in multiphase flows.
Based on the simulations conducted (Table 1-3), the following conclusions can be drawn, 1. For Low viscosity CFs: • R<1, and both svel are low: CF is slower than water, thus CF plugs can be seen.
• R<1, and both/only water's svel is high: CF plugs form but have a deformed shape as water tends to exert higher force on the interfaces of plugs.
• R>1, and both svel are medium/low: Since CF is faster than water, it is able to retain its shape and larger plugs are seen • R>1, and both svel are medium/high: intermittent long slugs that fill the entire pipeline can be seen.

For medium viscosity CFs:
• R<1, and both svel are low: larger plugs than previous case, as CF has higher viscosity and can retain shape better.
• R<1, but water svel is medium: Though higher water velocity will tend to deform the plugs, the increased viscosity of CF enables it to counter this near the wall where the velocity of water is lower. Thus, water successfully disperses CF finely in the entire medium, but at the walls a fine CF film is formed.
• R>1, and both svel are low: water plugs instead of CF plugs since water has the lower velocity.
• R>1, and both svel are medium/high: long slugs with wavy interface are formed.

For high viscosity CFs:
• R<1, and both svel are low: Due to high viscosity, CF interfaces remain intact closer to the walls. Thus, a coating of CF is constantly present at the wall. This is more pronounced at medium velocities of water • R<1, but both svel are medium/ high: Water disperses CF finely in the entire medium.
• R>1, and medium CF svel with high water velocity: thin film of CF coats the walls.
• R>1, and both svel are medium/high: a mix of dispersion and slugs with wavy interface are seen The understanding of how to apply these aforementioned observations will be discussed in the discussion section.

Generic Trends of Macroscopic parameters
Visual verification of flow regimes is not possible in sewage pipelines. Thus, it would be essential to monitor macroscopic parameters at various locations to verify the flow regime that is actually occurring. This section discusses the axial variation of pressure, CF fraction, and CF velocity.

Pressure
• It was observed that for a few lower viscosity cases, the injected CF does not retain its trajectory and eventually becomes a dispersed flow regime (e.g. L4 in Table 1). It was observed that at the location where the CF becomes dispersed, the pressure value in the pipeline dips towards the negative side, creating a vacuum pressure condition at the section (Fig 1a). • For cases L1 and H1, plots have been presented for comparison of variation of pressure along the central axis in the flow direction. (Fig 1 b,c). For the plug type flows as encountered in the case L1, the axial variation of pressure fluctuates suggesting a higher localized pressure due to the onset of plugs.

CF fraction and CF velocity
For many of the cases it could be observed that the trends for CF fraction and CF velocity were morphologically similar.
• As the flow regimes transformed to the dispersion patterns, the velocity variations were observed to attain a quasi-stable state with little fluctuations. This implies that for a uniformly dispersed flow regime, the velocity fluctuations can be minimum although the absolute magnitude of velocities may be higher.
• Similar plots can be observed for the flow regime comprising of wavy stratified along with dispersion of CF in water. Here, although there was a considerable oscillation (owing to the impact of wavy stratified structures) in the volume fraction plot, the CF velocity stood nearly a constant in the axial direction, even though the value was higher than the inlet boundary condition velocity value. (Fig 2. c, d) • Another significant observation was that, for higher inlet superficial velocities, the velocity values at different sections in the domain were reaching values higher than the inlet values. The larger velocities of CF plugs could be attributed to the lesser share of cross section available for flow leading to a nozzle like effect causing acceleration.
• Oscillatory nature of the volume fraction graphs with higher frequency and lower amplitude could also be correlated with bubbly flows while large amplitude small frequency oscillations of volume fraction graphs represented plug or slug flows. First and foremost, the present study has tested the CFD technique of Multi-fluid VOF and has demonstrated that it is a reliable tool that can be used to predict different flow regime formations, for multiphase horizontal flows in pipelines. This method has been shown to have better accuracy when compared with other popular interface capturing multiphase techniques whenever a wide range of flow regimes need to be predicted. This is a one of a kind simulation catering to a range of fluid viscosity ranging right from high to low.

DISCUSSION
The main result of our numerical simulations is that we are able to recommend superficial velocities at which cleaner-fluids need to be pumped into the sewage line, so as to achieve flow regimes that ensure delivery of cleaner fluid to the periphery of the tube walls. Our simulations indicate that most effective cleaning can be achieved by using a cleaner-fluid with a high viscosity ( ~5000 cP)). In such cases, a low-medium velocity (~0.05-0.3 m/s) of CF and water would ensure that the cleaner fluids are in constant contact with the pipe walls ( Fig. 3a). On the other hand, injecting cleaner fluids at very low superficial velocities (~0.03m/s) is not recommended as the thickness of the film at the wall becomes very thin (Fig.3b); consequently, this should be used only when the pipes are either new or in cases where cleaning cycles are usually very frequent.
When using medium viscosity CFs (~600 cP), the flow configuration as shown in Fig.3c, can be achieved; but such a flow regime can be achieved only when injecting cleaner-fluid at medium superficial velocity into a pipeline where waste water is flowing at a high superficial velocity. This is a very narrow restrictive range of operation. Moreover, this flow pattern would require large volumes of CF, since a lot of the cleaner fluid is also present in the core flow. This might be suitable for cleaning effluents from highly contaminated regions such as hospitals, where disinfecting the water in the core is also essential.

Fig 3: Flow pattern observed for (a) H-2 (b) H-1 (c) M-8 (d) L-3 (f) L-7. Red here is indicative of the cleaner fluid, while blue is indicative of the water effluent
When using low viscosity CFs (~16 cP), the best possible flow pattern that can be obtained is that of slug flow (Fig. 3d). This pattern is achievable when injecting cleaner-fluid at medium-high velocity into a pipe where the sewage water is moving at a low-medium velocity. In these cases, there is cyclic exposure of tube to waste water and cleaner-fluid. It also has to be pointed out that when using low viscosity cleaner -fluids, one must avoid injecting them at very low superficial velocities, especially when the waste water is flowing at a high velocity.
This might occur inadvertently in monsoon seasons when the sewage flow is quite large and the usual injection velocity will be too low in comparison. The effect of the large difference in the relative velocity causes the cleaner-fluid to completely disperse into the medium and does not provide effective cleaning (Fig. 3e).
Real life conditions might not adhere to expected behaviour. Thus there is a need to closely monitor the flow occurring in the pipeline. Since flow regime visualisation is not possible in pipelines, measuring devices need to be used at regular intervals along the length of the pipeline to measure the macroscopic parameters' variation.
The study also provides recommendations of what parameters to monitor on site to ensure adherence to expected patterns and can be applied in a wide range of conditions. Our numerical simulations show that pressure, cleaner with the aforementioned specific CF fluid property combinations of surface tension, viscosity and density, have been tested over a range of operating conditions and since the results are presented in terms of superficial velocities, these can be extended to pipelines of different sizes and different velocities. One way to apply the findings of this study, is by using chemical additives similar to the ones suggested above, to ensure that the chosen CF has properties similar to the ones used in this study. Alternatively, the simulation methodology used this study can be quickly used, with a given set of properties of the CF, and corresponding flow conditions that will ensure wetting of the walls by the CF, can be determined. Simulation using the method suggested in this paper will also greatly help in estimating flow behaviour when the geometry is scaled up and that too at a minimal cost. The validity of the theory discussed here can also be tested quite easily by conducting a preliminary test at any local municipal facility, house, building or hospital.
The importance of our result stems from the fact that it permits easy implementation in developing countries using minimal resources/equipment. This method also helps minimise cleaner fluid usage in order to restrict any environmental impacts. Our result have a wider range of applicability than just applying it to waste water treatment. Since, it is purely defined by fluid viscosity and fluid properties, it can be extended to address 'clean drinking water' problems. The solution proposed is very simple and can be implemented at every housing/ building and society level. Thus, our results offer an important new direction of research for interfacial hydrodynamics as a means for addressing different social problems at industrial levels. We offer an easy, affordable and readily implementable solution to curtail the outbreak of COVID 19. λ c is the friction factor vector given in the principal direction. l m is the length scale. µ = µ i α i + µ j α j , the mixture viscosity. The surface tension force is estimated as a Continuum Surface Force (CSF), following the work of [18]. An additional source term is appeared in the momentum equation on implementing this model for accommodating the surface tension force in the VOF calculations at the interface. Same/similar densities of bot

Multi
liquids-liquids since we want to prevent stratification based segregation. Other effects considered negligible in this case is wettability effects with the tube wall.

Procedure of Analysis
Shi (2015) [19] has considered a T-junction tubular domain with dual inlets-one each for the CF and water fluids (CF in the horizontal branch and water in the vertical branch). Each branch is having 200mm length and axial length extending up-to 8 meters downstream of the meeting junction. The simulation has been setup based on this experimental case with the validation carried out against the data presented in this paper. A 2D approximation was employed. A 2D domain was deemed sufficient since literature states that flow evolution is more dominated by events in the axial flow direction rather than the events in the transverse direction. The metering of most hydrodynamic parameters in the experiment being carried out in an 'averaged' approach, the 2D cases can present with equivalence for comparative validation without loss of generality. More Details of simulation setup has been provided in supplement.
The study has been performed for three different fluid systems consisting of three different CF viscosities-low, medium and high (Table 4). This is being done to compare the flow evolution features (including the regime of flow and other quantitative flow parameters like velocities of individual phases, etc.) as the CF viscosity varies.