Operating diagnostics on a flocculator-settling tank using Fluent CFD software

Transcription

1 Operating diagnostics on a flocculator-settling tank using Fluent CFD software S. Laine^ L. Phan<", P. Pellarin^, P. Robert^) (VAnjou Recherche, 1 place de Turenne, Saint Maurice Cedex, France (2)SOGEA, 3 cours Ferdinand de Lesseps, Rueil-Malmaison Cedex, France Abstract Recently, Computational fluid dynamics (CFD) software has become easy to use, fast and user-friendly. This new generation software offers an inexpensive means of testing and optimizing hydraulic operation of both existing constructions and those under design. In the present case, FLUENT CFD software was used to form an understanding of a periodic dysfunction at a drinking water production plant at a site in North Africa. The structure involved is a circular, 27-metre diameter, flocculator-settling tank, supplied with surface water taken from the nearby river. The following dysfunction was observed on the structure in question. In summer, during the hottest time of the day, a cloud of floes would sometimes rise to the surface, appearing in the centre of the settling zone. The cloud then progressed slowly to the outer zone of the tank, where a proportion of the total suspended solids would overflow, impairing the quality of the treated water. The phenomenon was easily linked to large-scale external temperature swings, sometimes rising as high as 50 C in the shade during the daytime. However, observations made in situ offered no explanation for the full range of mechanisms triggered by these temperature variations. The CFD simulations enabled to understand the precise mechanisms in operation when the phenomenon is observed and to have a good base from which to search for a solution to the problem.

2 260 Water Pollution 1 Introduction Recently, Computational fluid dynamics (CFD) software has become easy to use, fast and user-friendly. This new generation software package enables drinking water production station manufacturers to monitor and optimize the hydraulic operation of their constructions cost-effectively. In the present case, FLUENT CFD software was used by Anjou Recherche and SOGEA, both subsidiaries of VIVENDI (Generate des Eaux), to form an understanding of a periodic dysfunction at a drinking water production plant at a site in North Africa. 2 The problem The structure involved is a circular, 27-metre diameter, flocculator-settling tank, supplied with surface water taken from the nearby river. The raw water is injected with a coagulant into the centre of the tank, while the treated water is removed at the outer wall by overflow into a supernatant collection trough (see figure 1). The structure is divided into three zones by two vertical circular baffles. The narrow-diameter central zone is equipped with a powerful agitator that stirs up the raw water, causing suspended solids to coagulate. The wider secondary zone slows down the agitation movement and assists flocculation. The outer zone, which is even wider still, reduces agitation speeds to a point where a laminar regime is created, allowing the solids resulting from the flocculation process to settle. The sedimentation sludge from the settling process is scraped along the floor of the tank, towards a recovery gutter whence it is flushed out through the floor. The following dysfunction was observed on the structure in question: In summer, during the hottest time of the day, a cloud of floes would sometimes rise to the surface, appearing in the centre of the settling zone. The cloud then progressed slowly to the outer zone of the tank, where a proportion of the total suspended solids would overflow, impairing the quality of the treated water. Conversely, at all other times of the year, the facility operates perfectly well and supplies high-quality processed drinking water. 2.1 Methods The phenomenon was easily linked to large-scale external temperature swings, sometimes rising as high as 50 C in the shade during the daytime. However, observations made in situ offered no explanation for the full range of mechanisms triggered by these temperature variations. The complete process was finally elucidated thanks to the hydraulic simulation.

3 Water Pollution 261 Flocculation Zone Settling Zone Floccdator-settling tank Contours Fluent/UNS4.2 (axi, mcke, steady) Aiou Recherche - SOG^M 998 FluenMnc. Figure 1: Diametral view of the flocculator-settling tank Measures taken during days when the plant was dysfunction ing showed that the temperature of the feed water was sometimes 1.5 C higher than the temperature of water in the settling tank, which means that the river-water feeding the plant sometimes undergoes a heating process faster than the time needed for it to arrive to the settling zone. Sensing that the rise of the floes was directly due to this wide temperature differential, it was decided to study the structure's hydraulic behaviour before and during the inflow of the warmer water using the FLUENT CFD software, in order to come up with an explanation of the phenomena in operation. 2.2 Modelling theflocculator-settlingtank We modelled theflocculator-settlingtank as follows: - The structure is circular and symmetrical around its central axis. It was therefore modelled as a two-dimensional slice removed from the radius of the tank (see figure 2). - A square mesh was imposed on the slice so modelled, and this was tightened in the vicinity of the tank walls in order to take account of limit layer phenomena. - The agitation screw was modelled using a fan model, and the turbulence generated by the flow was computed using the RNG-ke method.

4 262 Water Pollution - The heat transfer equations were activated in such a way as to simulate conduction and convection phenomena. Water density was a function of its temperature. Symmetry axis Flocculator- Settling tank Grid Fluent/UNS 4.2 (axi. rrgke, Anjou Recherche - SdGEAH Fluent Inc. 2.3 Simulations carried out Figure 2: 2D axisymmetric mesh During periods of dysfunctioning during the summer, the water temperature in the settling tank stood at around 25 C, while the feed water temperature was higher, at 26.5 C. Two simulations were carried out: - The first was run in order to observe the plant's hydraulic operation before the warm water inflow. The flow was therefore computed in steady mode (timeindependent) using a constant temperature value of 25 C for the feed water and the water already in the tank. - The second simulation was carried out in order to visualise the disturbance caused by the arrival of the warm water. The flow was therefore computed in unsteady mode, using the previous set of results, and injecting water at a temperature of 26.5 C. In this case, the external air temperature, estimated at 45 C, was likely to heat the tank surface and was therefore also taken into account. Calculating the flow in unsteady mode enabled us to monitor the warm water's penetration into the structure over time, as well as the sun-induced water

5 Water Pollution 263 temperature increase, and changes to the hydraulic currents triggered by these events. 3 Results 3.1 First simulation The first simulation enabled us to check proper hydraulic functioning under normal operating conditions. This was because in steady mode, and at a constant 25 C temperature, three distinct agitation zones were clearly discernible, corresponding to the three normal functions (see figure 3): 1 - Coagulation = Rapid agitation; 2 - Flocculation = Slow agitation; 3 - Settling = Laminar flow. In the first, very turbulent zone, the agitation screw acted as an accelerator taking the water towards the flocculation zone. The water then passed from the second (flocculation) zone towards the third (settling) zone, moving along the floor of the structure until it reached the outer wall, where it moved back up to the surface. Because of this, the particles entering the settling zone are close to the bottom of the tank and rapidly settle onto it. This explains why the treated water recovered at the overflow when these conditions apply is of very good quality. Flocculator-settlinq tank Velocity Vectors Colored By Velocity Magnitude (nrvs) Fluent/UNS 4.2 (axi, rngke, steady) Anjou Recherche - SOGEA1998 Fluent Inc. Figure 3: Velocity vectors in steady mode at 25 C

6 264 Water Pollution 3.2 Second simulation The second simulation revealed the following phenomena (see figure 4, time = 15, 19, 20 and 21 mn): The warm water entering the structure penetrates rapidly the coagulation and flocculation zones, moving along the open surface. So far, the water is still heated by its contact with the air, before it sinks towards the base of the tank. The agitation maintained by the screw thereafter rapidly homogenizes the flocculation zone temperature at around 27,5 C. This water, now at a temperature of 27,5 C diffuses far more slowly through the settling zone, moving along the bottom of the tank, as already seen in steady mode. It was observed that the surface of the settling zone was also heated by the ambient air. However, due to its very slow circulation, the surface water was not mixing with the water in the bottom of the tank, whose temperature remained at a constant 25 C. For this reason, water heated to 27,5 C penetrating to the bottom of the settling zone has a lower density than that of the ambient water at 25 C and naturally rises, like a bubble, to the surface, before moving to the edges of the tank. 3.3 Flow lines By observing the flow lines over time, it was possible to understand how the trajectories of the fluid particles were deflected by the changes in temperature. - In steady flow mode (see figure 5, time = 0 mn), the water in the settling tank is moved by a single recirculation which sweeps its entire volume round. Water flowing in from the flocculation zone moves along the base of the tank until it encounters the outer wall, where it rises to the surface. Here, a fraction of the flow exits the tank by overflow, while the remaining fraction returns towards the centre of the tank moving along the upper surface until it reaches the second baffle, where it once again sinks towards the floor of the structure. - In unsteady mode (see figure 5, time = 17, 20 and 25 mn), as the water heated to 27,5 C penetrates the settling zone, a second recirculation is created around base part of the outer edges of the tank, moving in the opposite direction to the former. The creation of this second recirculation can be explained as follows: because of its lower density, the water coming from the flocculator tends to rise to the surface before it reaches the outer wall, the effect of which is to cut the existing recirculation in two. 3.4 Particle injection To verify that the changes observed in flows were causing the surface cloud of floes, we simulated the trajectories of particles of the same size and density as those present in the tank.

7 Water Pollution 265 v-4 Time = 15 i Y-4 Time = 0 mn Fbccifatof-s«ttir«ts Path Unes Colored by Cortoirso(SMi?TefTperatUB(c) Time = 19 mn v-4 Time = 17 mn Time = 20 mn Y- Time = 20 mn Time = 21 mn Time = 25 mn Figure 4: Contours of water temperature Figure 5: Flow lines

8 266 Water Pollution FlocoJator-settlingtank Particle Traces Diameter: 1 rrm - Density: 1.00 Fluent/UNS4.2(axi,.. _._... Ariou Recherche - SiOGEA 1! Fluent Inc. Figure 6: Particle injection in steady mode at 25 C Flocctlator-settling tank Particle Traces (TTrre=20rm1 Diamster:1rrm- Density:1.00 Ruent/UNS4.2(axi, Anjou Recherche - SOGEA1 Fluent Inc. Figure 7: Particle injection in unsteady mode

9 Water Pollution 267 It was thus observed that in steady mode at 25 C, particles move with the hydraulic currents and thus enter the settling zone moving along slightly above the base of the tank (see figure 6). Which means that they all settle onto the bottom plate rapidly, before reaching the outer edge of the tank, where the flow moves back up to the surface. On the other hand, when water at a temperature of 27.5 C flows into the settling tank, the bubble of warm water formed in the centre of the tank drags some of the lightest floes up with it before they have time to settle on the floor of the tank (see figure 7). These floes, which have been carried up to the surface, are subsequently slowly carried towards the outer collection trough. The phenomenon observed is thus entirely elucidated by the simulations carried out. 4 Conclusions The findings of hydraulic simulations carried out using FLUENT CFD software provided a full explanation of the rising clouds of floes sometimes observed in the flocculator-settling tank under study: - When the water temperature is constant, the tank operates in a perfectly satisfactory manner. The slow rate and the form of the flows in the tank provide a satisfactory settling rate of all the solids to be removed. - When the feed water is warmer than the surrounding water and the external air temperature rises to around 45 C in the daytime, the coagulation and flocculation zones heat up rapidly and homogeneously. Conversely, the settling area heats up more slowly and not in a more regular way. The wave of warmer water reaching the floor of the tank rises up in a bubble, carrying with it the lighter solids. These then appear on the surface of the tank as a cloud of floes, which temporally impairs the quality of the treated water. Thus, a CFD simulation enabled us to diagnose a periodic dysfunction of an existing plant, to understand the precise mechanisms in operation when the phenomenon is observed and to have a good base from which to search for a solution to the problem. This modelling is particularly important insofar as it is used to reproduce a phenomenon corresponding to the type of extreme conditions not usually taken into account at the design stage of water treatment plants. 5 Acknowledgements The authors would like to thank Miss Camille Arnaud, OTV technical collaboration in this study. engineer, for her

10 268 Water Pollution 6 References Adams, E.W. & Rodi, W., Modelling flow and mixing in sedimentation tanks, Hyd. Eng., Vol. 116, No. 7, pp , Fluent Inc., Fluent/UNS & Rampant Users Guide, Fluent, Imam, E., McCorquodale, LA. & Bewta, I.K., Numerical Modelling of Sedimentation Tanks, J. of Hydr. Eng., Vol. 109, No. 12, pp , Larock, B.E., Chun, W.K.C. & Schamber, D.R., Computation of sedimentation basin behaviour, Wat. Res., Vol. 17, No. 8, pp , Peerless, S.J., Basic Fluid Mechanics, Pergamon Press, Londres, Vallentine, H.R., Applied Hydrodynamics, 2^ edn, Butterworths Scientific Publications, Londres, 1969.