FUNDAMENTALS ON EQUILIBRIUM CONCENTRATION CURVES... 37 FUNDAMENTALS ON EQUILIBRIUM CONCENTRATION CURVES FOR SEDIMENTATION AND DIFFUSION OF PARTICLES IN FLUIDS Harry Edmar Schulz José Eduardo Alamy Filho Swami Marcondes Villela Department of Hydraulics and Sanitary Engineering, School of Engineering at São Carlos, University of São Paulo, Brazil (Convênio FINEP 23.01.0606.00/Laboratório de Hidráulica Ambiental-CRHEA/SHS/EESC/USP) Abstract Solutions for the equilibrium sediment concentration profiles, defined as the situation in which the settling movement of sediment particles is compensated by turbulent diffusion in channel flows, are presented and compared with the classical Rouse profile. In the Rouse solution, the settling velocity is considered independent of the sediment concentration, a very restrictive situation which leads to unrealistic results for the sediment concentration near the bottom of the flow. In this paper, the use of the mass conservation principle permits to obtain a more realistic solution for the equilibrium profiles. It is shown that it is possible to extend the obtained profiles until the bottom of the flow (which simplifies further calculations). Further, considering the result of the momentum equation (potential flow) around a sphere, a nonlinear approximation between the settling velocity and the volumetric concentration of sediment is obtained. It is discussed that nonlinear effects must be considered to obtain more realistic equilibrium profiles, exposing a situation very different to that of the Rouse profiles. The new profiles do not present the mentioned physical limitations. Key words: sediment concentration profiles, settling velocity, equilibrium condition. Introduction: The Traditional Low Concentration Case In the literature, the 2D steady-state turbulent advectiondiffusion equation is used to estimate the distribution of suspended sediment along fluid flows. The equation is given by: u ∂C ∂C ∂ ∂C +w = − c 'u ' + Dx ∂x ∂z ∂x ∂x ∂ ∂C ∂ − c ' w ' + ( w sC) + Dz ∂z ∂z ∂z (1) In equation (1) x and z are the longitudinal (horizontal) and vertical coordinates respectively [L]; u , w are longitudinal and vertical mean velocities [LT–1]; u ′, w ′ are the longitudinal and vertical velocity fluctuations [LT–1]; C is the mean suspended sediment concentration [ML–3]; c ′ is the fluctuation of suspended sediment concentration [ML– 3 ]; c ′u ′, c ′w′ are statistical correlations between concentration and velocity fluctuations [ML–2T–1]; Di is the molecular diffusion coefficient for the i direction (i = x,z) [L2T–1]; and ws is the sediment settling velocity [LT–1]. Equilibrium situations in channel flows imply that the concentration profile does not vary along the flow direction (longitudinal derivatives vanish, or ∂/∂x = 0). In such flows, mean vertical velocities can be neglected in comparison to the mean longitudinal velocity ( w = 0 ). This situation, a particular case for channel flows, is adopted as the initial condition for numerical simulations of complex flows. Zedler and Street (2001), for example, used the Rouse equilibrium profile to initialize an LES simulation for alluvial channels. The products c ′u ′, c ′w′ are usually quantified through the eddy diffusivity concept and assuming proportionality between momentum and mass transport, leading to equation (2) −c ' w ' = D tz ∂ C νt ∂ C = ∂ z σc ∂ z (2) Minerva, 1(1): 37-43 38 SCHULZ, ALAMY FILHO & VILLELA Dtz is the eddy diffusivity [L2T–1];ν t is the eddy viscosity [L2T–1] and σ c is an empirical proportionality constant (usually σ c = 1 to 0.74). It is known that Dtz >> Dz far from the channel bottom, so that Dz is neglected. For equilibrium situations, equations (1) and (2) permit to obtain the simplified equation (3): νt ∂ C + w sC = 0 σc ∂ z (3) For such simple flows, the eddy viscosity follows the parabolic distribution derived from the logarithmic velocity profile, as shown in equation (4): z ν t = κ ⋅ u * ⋅z ⋅ 1 − h (4) where κ is the von Kárman constant (0.4 for clear water); u* is the shear velocity [LT–1] and h is the flow depth [L]. Equations (3) and (4) lead to equation (5): ∂φ ∂Z =− σ Z (1 − Z ) ws ⋅ φ ⋅ c κ⋅u* (5) Exploring the High Concentration Case where φ is the nondimensional volumetric sediment concentration, c /ρs, which varies from 0 to 1; Z is the nondimensional distance to the bottom, z/h, also varying from 0 to 1; and ρs is the sediment density [ML–3]. Integrating this expression for constant ws lead to the Rouse equilibrium profile, widely diffused into literature, given by equation (6): φ a Z − 1 = ⋅ φa a − 1 Z L w0 Using the mass conservation law for settling particles In real cases, the settling velocity decreases while the particle approaches the bed, where higher particle concentrations are reached. Considering only vertical displacements, the integral mass conservation law for settling particles leads to a linear dependence between the settling velocity and the volumetric concentration. To obtain this equation, the schemes of Figures 1a and 1b may be followed. From the integral mass conservation equation we have: (6) where a is a reference level above the bed; φa is the reference concentration at Z = a; L is given by L = ∂c / (κ.u*) and wo is the constant settling velocity [LT–1], usually given by the Stokes solution for a settling sphere. Equation (6) shows that the evaluation of the volumetric concentration requires the reference value at some point (a) along the water depth. This point is generally assumed near the bottom, being a = 0.05, for example, a value adopted in the literature. The Rouse solution is a classical equation used in the study of suspended sediment in channel flows. Julien (1995) affirms that, despite of its shortcomings and simplifications, the Rouse expression conducts to qualitatively good results, Minerva, 1(1): 37-43 being considered accurate mainly for lower concentrations. Indeed, the hypothesis of constant settling velocities may be considered valid only for very low concentrations, where interactions among particles are ignored. According to Einstein & Chien (1955), Rouse equation predicts well concentration values for φ < 0.04, but deviates progressively for higher concentrations. In real flows, however, much higher concentrations are present, so that particle interactions cannot be neglected. Thus, the constant settling velocity is a very limiting hypothesis. An interesting consequence of this assumption is the prediction of infinite concentration values at Z → 0, a gross deviation from the daily observations. To overcome this problem, the Rouse expression is used only above the reference level. However, it also may show deviations close to the surface, where concentrations close to null are usually predicted. In this case, experimental results for fine particles studies show that turbulent mixing brings sediment closer to the water surface, resulting in somewhat higher concentration values in these regions. The purpose of this paper is to furnish better theoretical approaches for the equilibrium concentration profile, without some of the limitations observed in the Rouse profile. Physical considerations are followed to obtain better evaluations of the settling velocity. aV1 = AV2 or φ V1 = (1 − φ)V2 (7) It is known (see Schulz, 1990, for example) that a/A = φ, that is, the ratio between the areas defined in figure 1a equals the volumetric concentration. The relative velocity (wo) between a single spherical particle and the fluid is given by the Stokes law. In the present case, the following relation is straightforward: w 0 = V1 + V2 (8) From equations (7) and (8) it follows that: w s = w 0 (1 − φ ) (because V1 = ws) (9) FUNDAMENTALS ON EQUILIBRIUM CONCENTRATION CURVES... Using equation (9) into equation (5) and integrating the result leads to: φ = φa a Z − 1 a − 1 ⋅ Z L ⋅w o a Z − 1 1 − φa + φa ⋅ ⋅ a − 1 Z L⋅ w o (10) Equation (10) is still very elegant. It may be said that the Rouse equation (equation 6) is a simplification of this more general result. A difference between both predictions is the fact that the normalized profile φ/φa of equation (10) depends on φa, while equation (6) furnishes only one profile for any φa. However, both predictions become closer for φa?0. Concentration profiles for different values of φa and Lwo are presented in Figures 2a to 2d, using equations (6) and (10). In all cases, the reference level is a = 0.05. At the reference level (Z = a), equations (6) and (10) are adjusted to reproduce the reference concentration φa.. For points above the reference level, the concentration values calculated by equation (10) are higher than that provided by the Rouse equation. This indicates that the particle interaction effects propagate along all the flow depth. On the other hand, for points located under the reference level, equation (10) conducts to adequate finite values (φ ≤ 1.0), while equation (6) leads to unreal values tending to infinite for Z → 0. So, the profiles generated using equation (9) (the mass conservation principle for the particle movement) denotes that particle interactions decisively affect the sediment distribution along all the fluid depth and that this equation must be used to attain more realistic results. Figures 3 a and b illustrate the concentration profiles behavior given Figure 1 39 by equations (6) and (10) for different values of the product L.wo and φa = 0.8. It can be seen that for higher L.wo the concentration values are lower along the fluid depth (above the reference level a). In other words, the effect of the bottom propagates more intensively for lower L.wo values. This result is qualitatively coherent with the influence of the physical phenomena which take part on particles sedimentation. Lower L.wo values imply in: a) lower L values, related to higher u* values and, consequently, higher ressuspension of particles (or turbulent diffusion), or b) lower wo values, which imply in slow settling and, consequently, more time with the particles in the bulk liquid. An approximation using the momentum conservation equation (movement equation) for relative vertical motion between settling particles Although equation (9) incorporates the mass conservation principle for particle settling, it does not consider any relative motion among particles. In other words, all particles maintain the mean distance between each other during the downward movement. The information that, for example, one particle near the bottom can be at rest while others above it move trough the fluid was not considered. To verify how this condition can modify the settling velocity and the equilibrium concentration profile, some approximations are presented here. The sketch of Figure 4a is first considered, which permits to define φ. In this figure, Ro is the radium of a spherical particle and z is the radial distance with origin at the centre of the sphere. It is imposed that z corresponds to the equivalent radium of the volumetric region around the sphere which guarantees equation (11). (a) Ratio between the horizontal area occupied by particles and the total area of the flow; (b) sedimentation (V1) and upward flow (V2). Minerva, 1(1): 37-43 40 SCHULZ, ALAMY FILHO & VILLELA 4 3 π R 03 R0 3 φ= = 4 3 z πz 3 (11) Equation (11) may be understood as a definition of the volumetric density φ or as a definition of the distance z. Figure 4b shows a particle in rest while fluid and neighbor particles move in relation to it, mainly in the vertical direction. At any distance above the particle, the flow is retarded (the flow lines diverge). To obtain a first evaluation of this retarding effect on the flow and, consequently, on the neighbor particles, the solution of the potential motion about a sphere is used here (as described by Prandtl & Tietjens, 1934). Considering that a second particle at the position α z (Figure 4b) follows exactly the fluid movement (an ideal situation), the velocity of this second particle is given by equation (12). density (equations 14 and 15) than the linear equation presented before. The Stokes solution is still valid for low φ; while for high φ the zero velocity is attained “faster” than predicted by the linear model (equation 9). The present result is mainly theoretical, but many empirical studies in the literature show the settling velocity as a nonlinear function of the sediment concentration. Among them are, for example, Fischerström (1967), Merkel (1971a, b), Larsen (1977), Rölle (1990), Takacs et al. (1990), Krebs (1991), Zhou & McCorquodale (1992) and Baldock et al. (2004). Figure 5a shows a comparison among the different equations mentioned in this paper. Using equation (15) into equation (5) and integrating the result leads to: 1 Z= 1 1 1 1 1 − a φ 1 − φa L⋅w o L⋅w o 1−φ −1−φa 1+ e ⋅ ⋅ a 1 − φ φa (16) R w s = w P 1 − 0 α z 3 or 3 R0 w s = w P 1 − β z (12) α and β = 1/α3 are constants, related to the distance between two particles which guarantees equation (11). wP is the velocity “at infinitum”, that is, the velocity without the retarding effect. In this study, wp follows from equation (9), in the form: w P = w o (1 − φ ) (13) From equations (11), (12) and (13) we obtain equation (14): w s = w o [1 − φ][1 − βφ] (14) In the following discussion, a simplified case is considered, where α = β = 1. Equation (14) then simplifies to: w s = w o [1 − φ] 2 (15) It is necessary to stress that this is only an approximate solution. A particular value of α was used together with the direct application of the potential flow solution. However, an important result is that the equation for the settling velocity is a still more complex function of the volumetric Minerva, 1(1): 37-43 Once more, a simple equation is obtained. However, in this prediction, it is not possible to isolate the normalized profile φ/φa. Figure 5 presents the concentration behavior for φa = 0.8 and different values of Lwo. The comparison with Figure 3 shows that the nonlinear effect of the concentration on the settling velocity conduces to equilibrium profiles much more sensitive to the value of Lwo. So, such effects must be considered when the concept of equilibrium profile is used. Conclusions The Rouse model for equilibrium concentration profiles in channel flows leads to unreal values of concentration near the bottom of the flow (a volumetric concentration higher than unity constitutes a physically incoherent result). The constant settling velocity hypothesis is the factor that inhibits to attain better results. A more adequate alternative consists in the use of the mass conservation principle to calculate the settling velocity, which permits to obtain more realistic sediment concentrations along all the fluid depth. Considering the movement equation in the form of the solution of potential flow around a sphere, it was possible to show that a nonlinear dependence exists between the settling velocity and the volumetric sedimentation concentration. Equilibrium profiles based on the obtained nonlinear equation show that these profiles may be very sensitive to the settling velocity function (that is, its dependence on concentration). For procedures where the equilibrium profile concept is needed, it is suggested that the use of profiles based on constant settling velocities be avoided. At least the results based on the mass conservation principle must be considered. FUNDAMENTALS ON EQUILIBRIUM CONCENTRATION CURVES... 41 Figure 2 (a) Relative concentration profiles (φ/φa) for different φa values and Lw0 = 0.5; (b) usual nondimensional profiles for φa = 0.8 and Lw0 = 0.5; (c) relative concentration profiles (φ/φa) for different φa values and Lw0 = 1.0; (d) usual nondimensional profiles for φa = 0.8 and Lw0 = 1.0. 1 1 (a) Lw0 = 0.5 (b) 0.8 0.8 1.0 1.0 Z 0.6 Z 0.6 Lw0 = 0.5 0.4 1.5 0.4 1.5 0.2 2.0 0.2 2.0 0 5.0 0 5.0 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 Figure 3 Equilibrium profiles for φa = 0.8, a = 0.05 and different values of Lw0 (a) rouse profiles (equation 6); (b) equation (10), more sensitive to changes in the Lw0 value. Minerva, 1(1): 37-43 42 SCHULZ, ALAMY FILHO & VILLELA Figure 4 Constant velocity 1 1 (Rouse model) 0.8 0.8 0.6 0.6 Lw0 = 0.5 1.0 1.5 2.0 Z ws/w0 (a) Definition of φ and; (b) flow lines above a sphere in rest. Equation (9) 0.4 0.4 Equation (15) 0.2 5.0 0.2 (a) 0 0 0 0.2 0.4 0.6 0.8 1 (b) 0 0.2 0.4 0.6 0.8 1 Figure 5 (a) Different settling velocities used in this study; (b) equilibrium concentration profiles using equation (16) with φa = 0,8 at a = 0.05. Acknowledgements To FIPAI, FINEP, CAPES, FAPESP and CNPq, Brazilian research support institutions, which collaborate in different aspects of the maintenance of the studies of the research program on erosion, sedimentation and longtime reservoirs behavior. References BALDOCK, T. E.; TOMKINS, M. R.; NIELSEN, P.; HUGHES, M. G. Settling velocity of sediments at high concentrations. Coastal Engineering, v. 51, p. 91-100, 2004. EINSTEIN, H. A.; CHIEN, N. Effects of heavy sediment concentration near the bed on the velocity and sediment distribution. MRD Sediment Series, n. 8, U.S. Army Engrs Div., Missouri River, Corps of Engrs, Ohama, Nebraska, 1955. Minerva, 1(1): 37-43 FISCHERSTRÖM, C. N. H.; ISGARD, E.; LARSEN, I. Settling of activated sludge in horizontal settling tanks. ASCE Journal of the Sanitary Engineering Division, v. 93, SA3, p. 73-83, 1967. JULIEN, P. Y. Erosion and sedimentation. Cambridge: Cambridge Univ. Press, 1995. KREBS, P. Modellierung und verbesserung der strömung in nachklärbecken. 1991. Dr. Thesis presented at the Versuchsanstalt für Wasserbau, Hydrologie und Glaziologie (VAW) der ETH Zürich. Herausgegeben vom BUWAL, Bern, September. LARSEN, P. On the hydraulics of rectangular settling basins: experimental and theoretical studies. Department of Water Resources Engineering, Lund Institute of Technology, Report n. 1001, University of Lund, Sweden, 1977. 170 p. FUNDAMENTALS ON EQUILIBRIUM CONCENTRATION CURVES... 43 MERKEL, W. Untersuchungen über das verhalten das belebten schlammes im system belebungs-nachklärbecken. Gewässerschutz-Wasser-Abwasser n. 5, Institut für Siedlungswasserwirtschaft, TH Aachen, 1971a. SCHULZ, H. E. Investigação do mecanismo de reoxigenação da água em escoamento e sua correlação com o nível de turbulência junto à interface – II. 1990. Tese (Doutorado) – Escola de Engenharia de São Carlos, USP. MERKEL, W. Die bemessung horizontal durchströmter nachklärbecken von belebungsanlagen. GewässerschutzWasser-Abwasser, v. 112, n. 12, p. 596-600. Institut für Siedlungswasserwirtschaft, TH Aachen,1971b. TAKACS, I.; PATRY, G. G.; NOLASCO, D. A generalized dynamic model of the thickening/clarification process. Advances in Water Pollution Control, Proc. 5th IAWPRC Workshop on Instrumentation, Control and Automation of Water and Wastewater Treatment and Transport Systems, Yokohama and Kyoto, 1992. p. 487-494. PRANDTL, L.; TIETJENS, O. G. Fundamentals of hydro and aerodynamics. New York, U.S.A.: Dover Publications, Inc, 1934. RÖLLE, R. Der feststoffgehalt im rücklaufschlamm von nachklärbecken. Korrespondenz Abwasser, v. 37, n. 3, p. 260-269, 1990. ZEDLER, E. A.; STREET, R. L. Large-eddy simulation of sediment transport: currents over ripples. Journal of Hydraulic Engineering, ASCE, v. 127, n. 6, p. 444-452, 2001. ZHOU, S.; MCCORQUODALE, J. A. Modeling of Rectangular Settling Tanks. Journal of Hydraulic Engineering, ASCE , v. 18, n. 10, p. 1391-1405, 1992. Minerva, 1(1): 37-43 ffffffffffffffffffffffffffff
© Copyright 2026 Paperzz