An improved flocculator design for use in particle aggregation experiments

Transcription

1 Notes 723 MANN, IS. H Production and use of detritus in STRICKLAND, J. D. H., AND T. R. PARSONS A various freshwater, estuarine, and coastal marine eco- practical handbook of seawater analysis, 2nd ed. Bull. systems. Limnol. Oceanogr. 33: 9 1 O-930. Fish. Res. Bd. Can NAGATA, T Carbon and nitrogen content of nat- TENORE, K. R., AND D. L. RICE A review of trophic ural planktonic bacteria. Appl. Environ. Microbial. factors affecting secondary production of deposit- 52: feeders, p In K. R. Tenore and B. C. Coull RAUSCH, T The estimation of micro-algal protein [eds.], Marine benthic dynamics. Univ. South Carocontent and its meaning to the evaluation of algal lina. biomass. 1. Comparison of methods for extracting protein. Hydrobiologia 78: SASAKI, G. C., AND J. M. CAPUZZO Degradation of Artemia lipids under storage. J. Exp. Biochem. Physiol. 78B: SOKAL, R. R., AND F. J. ROHLF Biometry. Freeman. Submitted: 2 July 1992 Accepted: 10 August 1993 Amended: 17 September 1993 Limnol. Oceanogr., 39(3), 1994, , by the American Society of Limnology and Oceanography, Inc. An improved flocculator design for use in particle aggregation experiments Abstract-Experimental studies of aggregate formation usually require knowledge of the rate of fluid shear. A type of Couette device which consists of a fixed inner and a rotating outer cylinder provides an easily quantifiable two-dimensional laminar flow in the annular space between the cylinders. Early designs of such devices had several drawbacks, however, including the sedimentation of particles in vertically oriented devices and difficulties in the ease and repeatability of sampling in horizontal units. The new horizontal design presented here addresses these problems through the use of a movable endcap and sampling port. This device has been shown experimentally to minimize the effects of sedimentation and also to allow multiple samples to be taken without the inconveniences associated with previous horizontal units. Use of the device is illustrated by repeated measurements, over a period of 34 d, of the coagulation efficiency (a) of the diatom Skeletonema costatum. Floes are formed by aggregation of suspended particles. Theoretical and observational work has highlighted the importance of floes in particle dynamics in marine and freshwater Acknowledgments This research was supported by ONR contracts NO J and NO l-0226 awarded to H. G. Dam. The vertical Couette device was lent to us by James Bonner. We thank T. Kiorboe, B. Logan, G. Jackson, J. O Donnell, and D. Lee for their comments and suggestions throughout the study. The manuscript benefited from reviews by A. Alldredge, U. Riebesell, and an anonymous reviewer. We are particularly grateful to Joseph Bellantoni who made the drawings in Fig. 1. systems (Hill 1992; Weilenmann et al. 1989; Riebesell 199 1). Experimental work on sediment and phytoplankton floes (Edzwald et al. 1974; Kiorboe et al. 1990) has led to new insights into their formation and fate. Because fluid shear is postulated to be the main mechanism for floc formation (Jackson 1990), an important element in any experimental study of particle aggregation processes is a device capable of generating a defined fluid shear comparable to those found in aquatic environments. Shear rate, G (s-l), can be related to energy dissipation rate, c (cm2 s-~) and the kinematic viscosity, X (cm2 s-l), by the expression G = (E/X) ~. If one assumes a value of X = 10e2 cm2 s-l, a low value of c = lo-(j cm2 sp3 for oceanic environments (Soloviev et al. 1988) and a high value of E = 1 cm2 s3 for tidal channels (Grant et al. 1962), then shear rates would typically range from 10e2 to 10 SS. A device consisting of a fixed inner and a rotating outer cylinder provides an easily quantifiable two-dimensional laminar flow in the annular space between the cylinders. This instrument is known as a Couette device and the flow within it as Couette flow (Donnelly 199 1). The mean shear in the annulus is determined by the diameters of the inner and outer cylinders and the rotation rate of the outer cylinder (van Duuren 1968). Many of the early Couette devices with vertically ori-

2 724 Notes ented axes of rotation have been used for experimental studies of Couette flow itself (see Donnelly 199 1) or for flocculation experiments that incorporated settling processes (Hunt and Pandya 1984). Horizontally oriented devices have also been used for flocculation studies (van Duuren 1968; Kiorboe and Hansen 1993). Experimental studies of flocculation usually require frequent (every few minutes) sampling of the suspended particles in order to characterize the coagulation process (Edzwald et al. 1974; Kiorboe et al. 1990). We sought to design and build a Couette flocculator which satisfied the following constraints: the device must be easy to assemble, fill, and sample; repeated sampling must be possible with minimal flow disturbance; and particle concentrations and size distributions should remain homogeneous throughout the volume of the flocculator. While a vertical design would meet the first two requirements, the sedimentation of particles would not meet the third requirement (Hunt and Pandya 1984). Horizontal devices, although limiting the effects of sedimentation, usually present several difficulties in ease and repeatability of sampling. Previously, horizontal Couette flocculators required the container to be stopped from rotating and turned upright in order to discharge a sample (van Duuren 1968; Kisrboe and Hansen 1993). Once the sample was removed, that volume had to be replaced if the experiment was to be continued. This repeated dilution had to be included in all calculations and was an added source of error. At this point the container would be resealed, placed back in the horizontal position, and the rotation resumed until the next sample was to be taken. In this paper, we present a new horizontal flocculation device whose improvements include the ability to repeatedly sample the suspension without stopping, opening the unit, or diluting the original volume. The device is inexpensive, easy to operate, and has proven useful in studies of coagulation in marine phytoplankton. The Couette device shown in Fig. 1 was designed and built at the Marine Sciences Institute, University of Connecticut. The cylinders are made of clear acrylic tubing. The diameter of the outer cylinder is cm and that of the inner cylinder is cm. The working length of the device is -26 cm giving the an- nular space an initial volume of - 1.O liter (the length and hence, the volume decrease each time the device is sampled, see below). The inner cylinder remains fixed while the outer cylinder rotates during operation of the flocculator. This arrangement results in laminar shear flow in the annular space between the inner and outer cylinder. The horizontal-rotation axis limits the effects of sedimentation and eliminates many of the sampling difficulties through the use of a movable end-cap, made of nylon, which incorporates a sampling port. A sampi? is taken by inserting the sampling port into the annular space (the tube is inserted only during sampling, which requires < 10 s). As the tube is withdrawn, the end-cap is forced into the vessel by a few turns of the push rod, displacing the volume required to fill the sampling tube (7-8 ml). When retracted, the sampling tube is flush with the inner wall of the end-cap. The sample is then discharged from the sampling tube by means of an airfilled syringe which is connected to a secondary diagonal vent (not shown in Fig. 1) running from the outside of the end-cap to the sampling port. Gently depressing the syringe forces air through the sampling tube and releases the sample. Thus, the sampling tube remains empty between samples. This arrangement allows repeated sampling without stopping, reorientation, or opening of the container, and also makes dilution of the original volume unnecessary. Up to 10% of the original volume in the annular space can be sampled in the course of an experiment. Calculation of the mean shear rate, G, (s-l) in a Couette device is given by G, = (27rN/60)(2R,R2/R22 - R12). (1) N is rotations per minute, R, the radius of the inner cylinder, and R2 the radius of the outer cylinder (van Duuren 1968). In our flocculator, a variable-speed motor allows G, to be adjusted from 2.0 to 30.0 s-l. Tests were carried out with both the new device and a vertical Couette flocculator to examine the effects of particle sedimentation in the two units. The latter device has been described by Hunt and Pandya (1984). In the first set of tests, particles were suspended in 0.2-pm-filtered and microwave-sterilized (Keller et al. 1988) seawater. The suspensions were placed in both devices and samples taken

3 Notes 725 ROTRTING OUTER CYLINDER FlNNULAR SPACE O-RINGS DRIVE SHAFT w+ MOVABLE END-CAP I / Ill TRACTABLE 7 Y-1 SAMPLING PORT PUSH / ROD Fig. 1. Cutaway and outside views of new horizontal Couette flocculator. Diameters: inner cylinder = cm; outer cylinder = cm. Maximum volume of the annular space is 1 liter (see text). All major components are labeled. Secondary diagonal port and syringe referred to in text are not shown. at min intervals over 90-l 20 min. Samples were taken at the midpoints of both devices. Particle concentrations were measured with an ELZONE 280 PC particle counter equipped with a 120~pm orifice tube. Triplicate readings of each sample were done to account for counting error. The mean shear rate (G, = 2.4 s-l) was identical in each unit. One test used 5-pm latex beads. The other test used the diatom Thalassiosira weissjlogii. This species was chosen because it has been shown not to form aggregates in appreciable quantities (Logan et al. 1994; Kiorboe and Hansen 1993; our own unpubl. obs.) and would presumably behave similarly to the beads except for different sinking velocities based on Stokes law. In these experiments, samples were taken at the midpoint in the horizontal unit and at the top, middle, and bottom of the vertical device. In all cases, the slope of particle concentration vs. time was estimated and difference of the slope from zero was statistically tested (Sokal and Rohlf 198 1). The new device was then used in a series of experiments to examine the coagulation efficiency of another diatom, Skeletonema costatum. This species has been shown to form aggregates in laboratory experiments (Kiorboe et al. 1990). The experiments involved exposing a suspension of algal cells to a shear rate of 5.7 s- l and sampling the suspension over time. A bloom was simulated in a laboratory culture by inoculating f/2 medium (Guillard 1975) with S. costatum. Cell concentrations were monitored for - 1 month with a redilution of the culture with fresh medium after 3 weeks. Samples were taken from the culture on alternate days for coagulation experiments.

4 726 Notes 1.8x1 O4 y= x y= x 1 n top 1 y= x I 1.0x104~ ' ' ' ' ' ' ' ' ' ' Fig. 2. Concentrations of 5.0-pm latex beads in both the vertical and horizontal Couette devices. Error bars are + 1 SD around the mean. Some bars are smaller than points. The aggregation of cells in the flocculator would be indicated by a decrease in particle concentration accompanied by an increase in mean particle volume. The probability that algal cells will stick together upon collision (coagulation efficiency, stickiness, (x) can be calculated from a! = {slope7rexp[ 1.5(SD)2/d12]}/(7.S24@Gm). (2) Slope is the slope of a linear regression of In (particles ml-l) vs. time (s), SD the standard deviation of the initial size distribution of individual particles (pm), d, the initial mean particle diameter (pm), and 4 the initial particle concentration as volume fraction (ppm) (Ki0rboe et al. 1990). This model was developed for spherical particles and is used here for convenience. This model is similar to that of Edzwald et al. (1974) except for the correction for variation in initial particle diameter shown in the numerator in Eq. 2. This approach to estimating a! assumes that all particles move in straight lines until contact occurs between them (Smoluchowski s linear model) and that when particles collide and form aggregates the volume (4 in Eq. 2) is conserved (O Melia and Tiller 1993). The first assumption is correct for monodisperse suspensions but becomes less accurate as the size heterogeneity of suspensions increases (see Jackson and Lochman 1993 for examples dealing with cell chains and spines). The net effect of this assumption is to overestimate particle contact rates. The second assumption is correct early in the coagulation process but be Fig. 3. Cell concentrations of Thalassiosira weissflogii sampled at three levels in the vertical Couette flocculator (G, = 2.4 s-l). Error bars are + 1 SD around mean. Some bars are smaller than points. comes less accurate as aggregates are formed and their total volume is underestimated by neglecting to account for porosity of the aggregate. As coagulation proceeds, the effect of the above assumptions is to compensate each other (O Melia and Tiller 1993). Therefore, it is likely that the model used here to estimate cy is a good first approximation of the coagulation process. As expected, 5.O-pm latex beads sedimented in the vertical flocculator but not in the new horizontal device (Fig. 2); i.e. the particle concentration at the midpoint of the flocculator changed in the vertical unit (P < 0.025), but not in the horizontal unit (P > 0.75). When the same experiment was run with T. weissflogii (Fig. 3) the vertical unit was sampled at three discrete depths. Although the regressions were not statistically significant, the trends suggest decreasing cell concentrations at the surface (P < 0.20) and increasing at the bottom (P < 0.17), consistent with the particle sedimentation shown by Hunt and Pandya (1984) in the same vertical device. Further analysis of the data by ANCOVA (Sokal and Rohlf 198 1) showed a significant difference between the top and bottom mean cell concentrations (P < 0.01) and suggested (P < 0.08) that the slopes of the top and bottom regressions were different. Conversely, cell concentrations in the horizonal device again stayed relatively constant over time (Fig. 4), with no evidence of cell settling as indicated by the nonsignificance (P > 0.75) of the regression slope.

5 Notes x104-7 'E 4.0x104- y= x I 0 4, i T 1 3.5x x104 " " " " '1" " I 7 I Fig. 4. Cell concentrations of Thalassiosira weissflogii sampled from the horizontal Couette flocculator. Error bars are f 1 SD around mean. 8.0~10' The results of one of the coagulation efficiency experiments performed with the horizontal device and the diatom S. costatum are presented in Fig. 5. As expected for a suspension experiencing coagulation, particle concentration decreased and particle size increased with time. The coagulation efficiency, (Y, estimated from this experiment was This experiment was part of a >800-h study during which stickiness was measured repeatedly as the culture aged (Fig. 6). The coagulation efficiencies estimated from this series of experiments ( O. 13) covered a broader range than those previously reported for this species (0.05-O. 12, Kiorboe et al. 1990). This high variability in stickiness (a) values is not surprising, however, due to differences between our experimental techniques and those of Kiorboe et al. (1990). They ran their experiments under turbulent conditions and at higher shear rates (G, = 50 s-l) than those presented here (G, = 5.7 s-l). Kiorboe and Hansen (1993) have documented an inverse relationship between shear rates and measured coagulation efficiencies. Thus, it is difficult to make direct comparisons of LY between experiments run at different shear rates. Another significant difference in experimental techniques was that Kiorboe et al. (1990) used sonication to disrupt aggregates before experiments; we did not. Sonication may break up cells, releasing dissolved substances into the experimental medium. Coagulation efficiency may be dependent on the presence (or absence) of some dissolved extracellular compounds (O Melia and Tiller 1993; Kiorboe and Hansen Fig. 5. Results ofcoagulation efficiency experiment with Skeletonema costatum (G, = 5.7 ss*) showing typical change in particle concentration (A) and mean particle volume (B) as a function of time. Error bars are f 1 SD around mean. 1993). Thus, the lowest coagulation efficiency values (a! = , 0.002) shown in Fig. 6 may have been the result of diluting the culture medium (and any dissolved substances therein) immediately before these measurements. Subsequent long-term (> 300 min) experiments with the device described here have produced marine snow-sized (> 0.5 mm) floes from both S. costatum and Thalassiosira pseudonana resembling floes formed by transparent exopolymer particles (Alldredge et al. 1993). However, these large floes were very fragile and impossible to quantify. Most coagulation experiments have been done with vertical Couette devices (e.g. Edzwald et al. 1974; Weilenmann et al. 1989) or in devices incorporating a vertically oscillating grid to generate turbulence and to keep particles in suspension (Kisrboe et al. 1990). Both kinds of devices allow one to easily and repeatedly sample the suspension. However,

6 728 Notes 4 5 a TIME (hours) Fig. 6. Time-series observations for Skeletonema costatum showing repeated measures of stickiness coefficient (CY + SE, n = 7) as a function of culture age as represented by cell concentration and volume fraction (culture was diluted with fresh medium at 560 h). vertical Couette devices have the disadvantage of particle sedimentation during experiments lasting more than a few minutes. In the device with a vertically oscillating grid, one disadvantage is that the high shear rate (50 s-l) that results from having the grid oscillate fast enough to keep particles in suspension may result in aggregate breakup (Alldredge et al. 1990). Another disadvantage is that the shear is turbulent; hence, the flow is difficult to characterize. We have designed and built a horizontally oriented Couette flocculator which is a significant improvement over previous models. The ability to repeatedly sample the device without stopping, or opening the unit, or diluting the original volume makes this device appropriate for many kinds of particle aggregation experiments. Being a prototype, the new device could be improved. One problem is the occasional intrusion of small air bubbles into the annular space after the device has been started. This problem has been greatly reduced by tightening the tolerances between the end-cap and the inner and outer cylinder O-rings and careful maintenance and lubrication of these seals. We are also exploring the possibility of continuous sampling through the use of a peristaltic pump and a flow cell which would circulate the sample through a laser particle counter and return it to the flocculator (via a second port in the end-cap) without the need to reduce the volume of the flocculator. In this system, both the incurrent and outcurrent ports are flush with the end-cap and would eliminate the need for a sampling port. The inclusion of a laser particle counter into the system would allow us to measure and quantify the larger, more fragile floes we have observed in previous experiments. Another area which bears further investigation is that of secondary flow within the device. Preliminary analyses of the equations of motion for particles in a rotating cylinder when the flow is at steady state (J. O Donnell pers. comm.) suggest that the pressure field generated by the nonrotating end cap may cause a slow secondary circulation of particles between the inner and outer walls of the device in addition to the laminar flow around the main axis. There may also be a small area on the upward flow side of the inner cylinder where larger particles might be scavenged from the

7 Notes 729 general flow (G. Jackson pers. comm.); e.g. -4% of l OO-pm particles or - 1% of 50-pm particles will be lost over time from solution by hitting the walls of the cylinders (G. Jackson pers. comm.). The predicted loss rates, however, are small relative to those observed due to coagulation (> 20%) in Fig. 5A. Further, particle loss through the process described above would be manifested by a decrease in particle concentration but not an increase in particle size such as observed during coagulation (Fig. 5B). We did not observe these conditions during our experiments. Finally, experiments with both latex beads (Fig. 2) and cells of the nonflocculating diatom T. weissflogii (Fig. 4) showed no significant decrease in particle concentrations on the time scales in which we are interested. Thus, flocculator designs with dimensions similar to the ones presented here can be particularly useful to study the early stages of aggregate formation; i.e. when floes are small (< 100 pm in diameter). Overall, the device described here has proven reliable in a variety of experiments and is a useful tool in the examination of coagulation processes. Department of Marine Sciences University of Connecticut Groton Marine Sciences Institute University of Connecticut References David T. Drapeau Hans G. Dam Gary Grenier ALLDREDGE, A. L., T. C. GRANATA, C. C. GOTSCHALK, AND T. D. DICKEY The physical strength of marine snow and its implications for particle disaggregation in the ocean. Limnol. Oceanogr. 35: , U. PASSOW, AND B. E. LOGAN The abundance and significance of a class of large, transparent organic particles in the ocean. Deep-Sea Res. 40: 113 l DONNELLY, R. J Taylor-Couette flow: The early days. Phys. Today 44(11): To whom correspondence should be addressed. EDZWALD, J. K., J. B. UPCHURCH, AND C. 0. O'MELIA Coagulation in estuaries. Environ. Sci. Technol. 8: GRANT, H. L., R. W. STEWART, AND A. MOILLIET Turbulence spectra from a tidal channel. J. Fluid Mech. 2: GUILLARD, R. R. L Culture of phytoplankton for feeding marine invertebrates, p In W. L. Smith and M. H. Chanley [eds.], Culture of marine invertebrate animals. Plenum. HILL, P. S Reconciling aggregation theory with observed vertical fluxes following phytoplankton blooms. J. Geophys. Res. 97: HUNT, J. R., AND J. D. PANDYA Sewage sludge coagulation and settling in seawater. Environ. Sci. Technol. 18: 119-l 2 1. JACKSON, G. A A model of the formation of marine algal floes by physical coagulation processes. Deep- Sea Res. 37: ~ AND S. LOCHMAN Modeling coagulation ofalgae in marine ecosystems, p Zn J. Buffle and H. p. van Lceuwen [eds.], Environmental particles. V. 2. Lewis. KELLER, M.D., W. K.BELLOWS,AND R.R.L. GUILLARD Microwave treatment for sterilization of phytoplankton culture media. J. Exp. Mar. Biol. Ecol. 117: KIBRBOE, T., K. P. ANDERSEN, AND H. G. DAM Coagulation efficiency and aggregate formation in marine phytoplankton. Mar. Biol. 107: AND J. L. S. HANSEN Phytoplankton aggregate formation: Observations of patterns and mechanisms of cell sticking and the significance of exopolymeric material. J. Plankton Res. 15: LOGAN, B. E., U. PASSOW, AND A. ALLDREDGE Variable retention of diatoms on screens during size separations. Limnol. Oceanogr. 39: O'MELIA, C. R., AND C. L. TILLER Physicochemical aggregation and deposition in aquatic environments, p In J. Buffle and H. p. van Leeuwen [eds.], Environmental particles. V. 2. Lewis. RIEBESELL, U Particle aggregation during a diatom bloom. 1. Physical aspects. Mar. Ecol. Prog. Ser. 69: SOKAL, R. R., AND F. J. ROHLF Biometry: The principle and practice of statistics in biological research. Freeman. SOLOVIEV, A. V., N. V. VERSHINSKY, AND V. A. BEZ- VERCHNII Small scale turbulence measurements in the thin surface layer of the ocean. Deep- Sea Res. 35: VAN DUUREN, F. A Defined velocity gradient model flocculator. J. Sanit. Eng. Div., Proc. Am. Sot. Civ. Eng. 94: WEILENMANN, U., C. R. O MELIA, AND W. STUMM Particle transport in lakes: Models and measurements. Limnol. Oceanogr. 34: Submitted: 28 April 1993 Accepted: 10 November 1993 Amended: 7 December 1993