ABSTRACT INTRODUCTION

Transcription

1 A new aspect for evaluating the effects of the tank hydraulic characteristics and wastewater quality on the overall efficiency of a liquid solid separation system J. Mihopulos, H.H. Hahn Institut fur Siedlungswasserwirtschaft (Sanitary Engineering), University of Karlsruhe, Germany ABSTRACT Traditionally chemical dosing for an improvement of liquid-solid separation is designed and operated more or less independently of the actual reactor's geometry and hydraulic characteristics. The wide-spread use of jar tests in day-to-day operations illustrates this fact. There is not much information available on the quality of floes emanating from coagulation units relative to the separation process' requirements. And if such information were available the interaction of suspensa with the flow structure in any given reactor still needs to be determined. The paper describes the results of systematic investigations on the interaction of specifically coagulated suspensions (under defined chemical and physical boundary conditions) with various separation reactors of different but defined geometry. It is intended to formulate correlations between separation efficiency, hydraulic characteristics (of a separation unit through tracer studies) and suspension properties as expressed in terms of reactor theory. INTRODUCTION Liquid-solid separation processes are considered as the main operations in water and wastewater treatment plans. In practice settling / flotating column tests are recommended for the rational design of separation tanks. Graphical methods to determinate the "suspended solids removal - surface overflow rate" relation are cumbersome and sometimes misleading in a very limited data region. One is needed to know the chemical characteristics of the applied suspension as well as the flow boundary conditions of the specific operation tank. A flow simulation of a separation tank is a difficult problem involving complex boundary conditions, density currents, two/three-phase and normally non- Stokes flow. The study of such flow behaviors has aroused considerable interest in recent years as indicated in a recent review by Stamou and Rodi (1985) [1]. In a real basin most of the kinetic energy of the flow entering the reactor is lost through eddies in the inlet zone. These result from the sudden expansion of the flow in the inlet section. Such phenomena cause generally recirculation and dead spaces.

2 472 Water Pollution The main objectives of this research was to experimentally study the effect of geometry on the hydraulics of separation tanks in the absence of density currents using model suspended solids. Sedimentation and flotation reactions have been investigated in rectangular and circular basins with moderate to significant depth. Such possible interactions are described schematically in Figure 1. It can be seen that a rapidly aggregating suspension leads to very efficient separation in (nearly) all types of separation units. Contrary to this a less readily aggregating system can only be handled effectively in separators of optimum design. A method developed in this report will be useful to evaluate the relative superiority of hydraulic tank design in a given chemical environment (special properties of the contaminant) in comparison to any other batch operation procedure. THEORY Unknown interaction of hydraulic characteristics and separation efficiency In an ideal liquid solid separation (sedimentation or flotation) tank, fluid travels smoothly from inlet to outlet without lateral dispersion. Fluid particles move in parallel paths with plug-flow or piston flow pattern (fig. 2). A given quantity of tracer added (pulse input technique) in a short interval of time at the influent end will travel without lateral dispersion and appear at the exit after an interval of time equal to the theoretical detention time. Ideally, there are neither dead zones nor short-circuiting. The flow-through curve (also called C-curve), defining the concentration of the tracer in the effluent as a function of time, has a standard deviation as well as variance equal to zero, because there is only one observation. Obviously, this ideal tank does not exist in practice. The hydraulic flow pattern in a real tank may be called dispersed-plug flow with dead spaces lateral dispersion and short-circuiting. The more the deviation of a real tank from the ideal, the lower the removal efficiency of the separation system. The C-curves for a group of real settling tanks are shown in figure 3. Tank No. 1 is a better design than tank No. 2, and tank No. 2 is a better design that tank No. 3. The coordinates of the dimensionless C-curves are / and -~r, C where: C : concentration of the tracer at the effluent after time t, CQ : total weight of tracer added at the influent divided by the volume of the tank T : theoretical detention time

3 Water Pollution 473 The most important purpose of conducting a dye test for a separation unit is to select a representative parameter which will be correlated closely with solids removal efficiency of the tank. There are from literature [2] different parameters already known and which are based on the following observations: time for 10 percent of the injected tracer to pass at the effluent end time for 90 percent of the injected tracer to pass at the effluent end (U, -r as Morril index, and /io short circuiting indices Each index or observation shown above is, in a way, a good indicator of the expected performance of a liquid-solid separation system, however, a small change in the tracer injection technique results in significant variation in these observations. Several investigators [3,4] have indicated the difficulty in reproducing these values after repeating the tests under identical conditions. In the opinion of the author, a statistically sounder index with good reproducibility is needed from C-curves to: + improve the efficiency of a separating system, + point out the undesirable elements in a tank, and + evaluate the relative characteristics of a group of sedimentation or flotation tanks designed by different manufacturers The hydraulic characteristic best suited for the purpose seems to be the so called "dispersion number" as an index commonly used in chemical engineering literature: d- where D : axial dispersion, u : mean flow velocity and L : length of travel The dispersion number is calculated from the variance of the C-curve. Variance determination represents and includes all points in the curve; whereas the other parameters include only one or two points in the curve. Hence the dispersion number seems to be a better indicator of degree of deviation from ideality; this number or diffusivity coefficient is calculated as follows: and _2_ r "

4 474 Water Pollution where <j^ : variance of C-curve in <j^ : variance of C-curve, time units dimensionless Levenspiel [3] has suggested three different equations relating the dispersion index with variance for open vessels, closed vessels and the combination of the above. For this study the closed vessel variance is selected: cr = Zd - Zcr 1 - e Unknown interaction of aggregation and separation efficiency As has become apparent from a quick view of the most literature sources, the design engineer needs to know - in principle - (a) the hydraulic loading/throughput and (b) the suspension characteristics such as particle or floe size, flocculation tendency, density etc. In general it is thought that it's optimal for a sedimentation tank to make solid aggregates large and heavy. And synonymous with this a flotation apparatus is thought to work best with light floes. On the basis of coagulation and flocculation theory on one hand and a hydrodynamic analysis of separation tanks on the other hand one can formulate hypotheses on floe and suspension properties. An analogy between sedimentation floe requirements and flotation requirements is from Hahn [5] already recognized. There are also significant differences, for instance in the desirable floe density or hydrophobicity. It is also known that the gas phase, the third phase in flotation, is responsible for significant differences between sedimentation and flotation due to its interaction with the solid phase. But which universal floe property can be directly correlated to the separation efficiency of a sedimentation or flotation process independently from the flow conditions? In order to clarify this one would have to perform detailed investigations in "microscopic dimensions" (i.e. single floes, defined hydromechanic conditions, etc.). Such experiments for systematically varied boundary conditions will allow to quantify the requirements for floes to be separated in specific liquid-solid separation tanks. For this study is an old parameter with the separation efficiency correlated: the collision efficiency number a. This parameter represents the process efficiency during orthokinetic flocculation; = -a x the measure of the degree of destabilization of the suspension. The above function is based on a relatively simplistic view (old rectilinear model) [6,7] of two particle collisions, in that they ignore the effects of the changes in fluid motion and short-range forces as the particles come to one another. The problem arise

5 Water Pollution 475 when the one particle is relative large compared to the other. The old model overpredict such collisions. Law/er et al [8] proposed a new model (curvilinear collision frequency functions) in which the collision factors are properly corrected. In the presented investigations, the prediction of the a for various (types of) chemicals has been made by using monodisperse suspension (Latex of 10 urn - polystyrene) according to the rectilinear model; the curvilinear model is used to correct the case of heterodisperse suspensions. The flocculation activity of the chemicals is measured and the chemicals are classified using the factor a. The same results reported in the literature [9] showed by using model suspension (micro-glass spheres 3-5 jam). Experimental Set-Up Three sedimentation tanks (two rectangular: a short-tank A and a long-tank B as well as one circular-tank C, with axial feed) and one rectangular flotation tank, geometrically similar to tank A (with the principle of electrolytic bubble production for easy control of bubble generation or Air/Solids ratio) were selected to demonstrate the relative superiority of using the dispersion number and the collision efficiency factor, to correlate separation efficiencies, over other parameters. The experiments were basically of two types: the tracer studies with tap water and the separation studies with model particles - described above - suspended in tap water. Details referring to the nature of suspending medium, suspension and coagulation / flocculation conditions or other experimental setup data are not presented here because of the limited extension of the paper. Tracer studies: Methyl-blue was selected as a tracer and a fluorometer was used to determinate the concentration of the dye in the effluent. The desired hydraulic flow rate into the tank was maintained initially for about five to six minutes - this method is recommended by Tekippe and Cleasby [2]- after which a drain tube was used to add the dye. The exact manner of adding the tracer as well as the mathematical model used after, for the data interpretation, must be standardized carefully to obtain reliable results. Actual mean detention time, variance of the C-kurve, and dispersion indices were determinated for each test by means of a computer program. Separation studies: The system had to be planned and maintained carefully to ascertain constant concentration of 100 mg/1 suspension in the supply tank. The various shapes of the tanks were employed as well as the experimental setup including pumps, mixers, flocculation reactors etc., are shown in figure 4. Concentrations and floe size distributions at the influent and effluent are continuously measured by using a particle analyzer CIS-I [10]. Low flow experiments for sedimentation (1 m/h) and fast flow experiments for the flotation (6 m/h) were conducted for average detention times of about 40 and 10 min.

6 476 Water Pollution respectively. As expected, steady state effluent concentrations were obtained after initial unsteady state (short interval of time). The mean concentrations of suspended solids in the influent and effluent streams at the time of steady state period were used for calculating the solids removal efficiency of each tank at a given overflow rate. RESULTS AND DISCUSSION The variance of C-kurve and the dispersion number (or index) of the tank seem to be representative, consistent and reproducible parameters because they consider all the points in the C-kurve rather than only one or two points as used in other parameters. To substantiate this statement, six selected tests were repeated under identical conditions of flow rates. Two pairs of tests per sedimentation tank showed that the dispersion number was reasonably reproducible. Tracer studies with the flotation tank were not compared to the sedimentation one because of the different models were used to interpretate the data. Although flotation column tracer studies existing in the literature [11], there is lack of information in terms of continuous flotation plans in which, tracer studies data require a complicated multiparameter model, and are not based on the Levenspiel's [3] dispersion model. Further studies are being planned to evaluate possible dispersions in the flotation cell in a continuous separation sytem. Another interesting observation was that dispersion number increased gradually when overflow rate was increased, for almost all the tank types (fig. 5). For a given overflow rate, the value of d was less for the long tank B than the A. In other words tank B exhibited characteristics which were closer to the ideal plug flow than the tanks A and C The circular tank shows a more stable hydraulic situation with d-numbers increasing slightly along the x-axis. This behavior ties in with the solids removal efficiency. The tank with the lower value of dispersion is more efficient than a tank with larger dispersion, independent of the chemicals were used to coagulate / flocculate (fig. 6). For a given overflow rate, tank B is more efficient than A. The relative superiority of tank B over tank A is suggested by the tracer studies and proved by the sedimentation studies. For the tank C one could think to be an exception to this rule. Tank C with relative high d-numbers showed a really good separation efficiency. A thoroughly evaluation of the results shows that not exceptionally the separation stability of the tank is also suggested from its hydraulic performance. This tank is an upward clarifier with completely different flow structure than tanks A and B (horizontal flow).

7 Water Pollution 477 A review of the literature [12] indicates that certain investigators have confined their work only to tracer studies and others have restricted that only to flocculation - separation studies. The present report is of a kind, where data and conclusions from tracer tests have been interpreted meaningfully and proved with separation studies under different chemical boundary conditions. Comparing at the other hand sedimentation and flotation one can recognize, that for identical flow structures - upward clarification as in flotation tank or sedimentation tank C - there are similar separation efficiencies reported (fig. 7) for not readily coagulated systems such as AP* or Fe*" floes. Increasing the Air- /Solids ratio - that means introducing more bubbles to the system - become the flotation more competitive. Readily coagulated and flocculated suspensions (by using polymers) appear to be removed with higher efficiency in flotation units. The reason: heterogeneous size distribution floes are favored in flotation units depending on the relative bubble (with various size distribution) attachment. These effects are also reported by other authors (Rosen [13]). According to the results one of the practical applications of this investigation will be to evaluate the tank performance requirements in terms of understanding the hydraulic flow conditions and the used suspension characteristics as well; and to relate sedimentation with flotation efficiencies in a meanfull way. For example, two tanks of different geometry designed by two equipment manufacturers can be subjected to dye tests for calculating the hydraulic flow characteristics (i.e. direction) and the dispersion index as well. For a given overflow rate and under identical chemical boundary conditions the tank with lower dispersion values can be judged to be a better design for the specific wastewater, to be removed. An improvement of the tank design, based on dispersion number measurements or any other mathematical combination of this criterion, is also possible. CONCLUSIONS Parameters like dispersion number which include all observations in the C- curve are more reliable than any one or two observations in the same curve Dispersion measurements seem to be a good design characteristic, especially when it's combined with the coagulation or flocculation collision efficiency test of a given suspension in terms of factor a Previous investigations have indicated difficulties in reproducing single-point observations (in tracer studies) and to correlate these data with solid removal efficiencies. At this study is such a method proposed. The disadvantage of

8 478 Water Pollution the suggested method may be that the tracer tests must be conducted under very carefully maintained constant environmental conditions > Observations based on the suspension characteristics have shown that: + there are separation units that show good efficiency even for less effectively coagulating or coagulated systems (i.e. Al^), while there are other types of sedimentation tank designs that show unsatisfactory performance; + for some types of reactor design the effectiveness of the coagulation process is more significant than for others (i.e. tanks A, B); some other sedimentation tank designs are more or less independent of the effectivity of the preceding coagulation stage (i.e. tank C); + the efficiency of a flotation unit of a geometry similar to that one of a sedimentation tank is noticeably higher for badly sedimentating floes - indicating a possibility for upgrading inefficient sedimentation units; > Design formulae based on quiescent settling or floating column analysis (like "jar tests") may not be as accurate as the formulae based on continuous-flow laboratory model studies. However such studies are tedious and require more and accurately planning and work FURTHER STUDIES One of the unanswered questions after this report will be to demonstrate three dimensional relations like these already demonstrated in figure 1. Mathematical models or diagrams are needed to evaluate these relationships; big scale experimental investigations are necessary to correlate the results predicted from the models and these produced from the experiments. New parameters could in this manner developed for a more accurately design of liquid-solid separation tanks. ACKNOWLEDGMENTS This investigation was a part of a project, which has been supported by the German Research Agency called (Deutsche Forschungsgemeinschaft). The authors wishes to acknowledge the cooperation of the students (of the department of Siedlungswasserwirtschaft) for their help on this project. The views expressed in

9 Water Pollution 479 this article are those of the authors only and are not necessarily endorsed by either of these agencies or persons. REFERENCES 1. Stamou A. I. and Rodi W. "Review of Experimental Studies on Sedimentation Tanks", SFB 210/E/2, Universitat Karlsruhe, SFB -Verlag, Karlsruhe, Tekippe R. J and Cleasby J. L. "Model Studies of a Peripheral Feed Settling Tank", J. San. Eng. Div. Proc. Amer. Soc. Civil Engr, Vol. 91, SA1, pp. 85, Levenspiel O "Chemical Reaction Engineering", John Wiley and Sons, Inc., Chapter 9, New York, Al-Khafaj A. A. A. R "Effect of Inlet Design on Efficiency of Center-Feed Sedimentation Tank Models", M.S. Thesis, Iowa State University of Science and Technology, Ames, Iowa, Hahn H. H. "Quo Vandis Chemical Treatment?", (ed. Hahn H. H. and Klute R - Springer Verlag), pp. 1-14, Proceedings of the 4th Gothenburg Symposium on Chemical Treatment, Madrid, Spain, October 1-3, Camp T R and Stein P. C "Velocity Gradients and Internal Work in Fluid Motion", J of Boston Soc. Civil Engrs., 30:219, Smoluchowski M. "Versuch Einer Mathematischen Theorie der Koagula- 7", Z. (Jour.). Physik. Chem., 92:129, Mooyoung H. and Lawler D. F. "The (Relative) Significance of G in Flocculation", J AWWA, pp , October Mihopulos J. and Hahn H. H. "The Effectively of a Liquid-Solid Separation System...", (ed. Hahn H H and Klute R. - Springer Verlag), pp , Proceedings of the 5th Gothenburg Symposium on Chemical Treatment II, Nice, France, September 28-30, CIS-I, (Computerized Inspection System I), "Manual", LOT GmbH, (former ORIEL), Im Tiefen See 58, D-6100, Darmstadt, Rajinder P. and Masliyah J "Flow Characterization of a Flotation Column", The Can. Jour. Chem. Engr., Vol. 67, pp , December Villemonte, J. R, Rohlich G. A. and Wallace A. T "Hydraulic and Removal Efficiencies in Sedimentation Basins", In "Advances in Water Pollution Research", Proceedings of 3rd Conference on Water Poll. Res, Vol. 2, pp. 381, Munich, Germany, Rosen B. "Dissolved Air Flotation", (ed. Grohmann A., Hahn H H and Klute R - Gustav Fischer - Verlag), pp , Proceedings of the 1st Gothenburg Symposium on Chemical Water and Wastewater Treatment, Gothenburg, Sweden, 1984

10 480 Water Pollution FIGURES Overall efficiency Wastewater quality Figure 1: Schematic representation of the effect of wastewater characteristics and reactor geometry upon the overall efficiency of liquid-solid separation Figure 2: Fluid particles move in parallel paths with plug flow or piston flow pattern

11 Water Pollution 481 TRACER IS ADDED AT THE INFLUENT END AT 1*0 2.0 Figure 3: The flow-through curve (C-curve) defines the concentration of the tracer in the effluent as a function of time EXPERIMENTAL SET-UP (Engineering flow diagram) nch.m,c.l Slor,«, Tank Figure 4: Experimental Setup (Engineering Flow Diagram)

12 482 Water Pollution Vessel dispersion number D/u*L 0,2 Tank A B Temperature (In-Out) <= 0,2 C 0,15-0,1 0,05 Hydraulic boundary conditions for the investigations 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1 Overflow rate m/h Figure 5: Tracer test results indicate that the dispersion number increased as overflow rate increased Sedknmntatlon. F#(3+)+Polym*r Y - A «EXMB * X) + C O.4 O.6 O.8 1.O Overflow rate (m/h) Figure 6: Solids removal efficiencies decreased as the overflow rate increased and (as the dispersion number increased-compare with fig. 5)

13 Water Pollution 483 SEPARATION PROCESSES DIAGRAM Sedimentation(1 m/h) - Flotation(6 m/h) Removal efficiency 0,8-0,6-0,2 / 0 0,1 /""" without x \ coagulant / 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 Collision efficiency factor (a) "paddle" Figure 7: Observed interaction of chemical suspension parameters and physical reactor characteristics in the removal as solids in a series of sedimentation/flotation tanks