HARVESTING SOME AGRICULTURALLY PROMISING ALGAE ' Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY JOHN BENHAM GERRISH ’ 19:22 " This is to certify that the thesis entitled Harvesting Some Agriculturally Promising Algae presented by John Benham Gerrish has been accepted towards fulfillment of the requirements for __Eh.D_.__degree in Agricultural Engineering g a 2'2 g% %/5 Major professor Date .NoxemheLlfl._1912 IIIIIIIIIIIIIIIIIIIIIIIIIII 72.--,w] 07 2 Mat 4' _’ 5 {wk 2‘? m f?“ "‘ ”'35“ l. ‘3 mes; ’~'~. 'r' 15”. I ABSTRACT HARVESTING SOME AGRICULTURALLY PROMISING ALGAE By John Benham Gerrish Several microscopic algae were harvested by each of several methods. Combinations of species and method were sought which might make algal culture an attractive alternative for a limited-resource agricultural enterprise. Genera represented are Chlorella, Oocystis, Scenedesmus, Coelastrum, Chlamydomonas, Anabaena and Spirulina. The algae were produced in light-limited lO-liter continuous culture. Harvesting methods investigated include sedi— mentation, centrifugation, electroflocculation, filtration, electrodecantation and predation. Quantitative measure of harvesting success is attempted and a new parameter based on entropy is presented. A stochastic model for sedimen- tation provides an incomplete description of the process. There are'significant differences in harvesting behavior between cells grown at different growth rates in the light- limited cultures; old cells harvest generally more easily than young cells. Energy costs are estimated for acceler- ting the harvest operation. DeSartment C airman HARVESTING SOME AGRICULTURALLY PROMISING ALGAE By John Benham Gerrish A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Engineering 1972 To Sandra, and our four children: Deborah Philip Elizabeth Matthew ii ACKNOWLEDGEMENTS I am grateful to the many people here at Michigan State University who so graciously allowed me on occasion to pick their brains. Their names are too many to list here, but I will try to thank them by giving advice in the same generous spirit shown me. To members of my doctoral guidance committee go special thanks: Dr. W. G. Bickert and Dr. J. B. Holtman, Department of Agricultural Engineering, Dr. K. L. Schulze, Department of Civil and Sanitary Engineering, and Dr. C. P. Wolk, Department of Botany and Plant Pathology. They have treated me as a friend and colleague. Dr. Bickert dili- gently served as my major professor in a venture which was quite foreign to his background. I thank him for the freedom he provided me in his own willingness to learn. Dr. C. W. Hall, former chairman of the Department of Agricultural Engineering saw to it that I had all the equipment which I needed and provided some encouragement besides. Dr. B. A. Stout, current chairman of the depart- ment, has accepted deadline postponements and yet managed to smile on my research efforts in the benign way depart- ment chairmen sometimes smile. Thanks. iii I also thank Miss C. Greenamyre, Mr. J. P. Harper, Miss M. M. Hof and Mr. W. S. McAfee for their part in the completion of this thesis. Miss Greenamyre served as a very observant laboratory assistant; Mr. Harper served as Devil's Advocate in a very helpful way; Miss Hof prepared an elegant first draft from my illegible mutterings; and Mr. McAfee helped with the scanning electron microscopy. My father— and mother—in-law, Mr. and Mrs. Folmar Bjerre provided a month's refuge for my wife and four children while I assumed the monkish attitude required to get this thesis written. Thank you. iv i 0.. s... TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . . . . . vii LIST OF FIGURES . . . . . . . . . . . . . . . . . . ix GLOSSARY OF SYMBOLS . . . . . . . . . . . . . . . . Xi CHAPTER I. INTRODUCTION . . . . . . . . . . . . . . . . 1 II. OBJECTIVES . . . . . . . . . . . . . . . . . 5 III. LITERATURE REVIEW . . . . . . . . . . . . . . 9 IV. EXPERIMENTATION . . . . . . . . . . . . . . . 33 Algal Species . . . . . . . . . . . . . . . 33 Harvest Methods . . . . . . . . . . . . . . ”0 Materials . . . . . . . . . . . . . . . . . 51 Species . . . . . . . . . . . . . . . . . 51 Media . . . . . . . . . . . . . . . . . 51 Equipment . . . . . . . . . . . . . . . . 67 Procedures . . . . . . . . . . . . . . . 83 Measuring Techniques . . . . . . . . . . 83 Management of the Culture . . . . . . . . 85 Procedures Followed in Harvesting the Algae . . . . . . . . . . . . . . 98 Measurement of Success . . . . . . . . . . 102 Prediction of the Behavior of Recovery R with Time in a Batch Settling Operation . . . . . . . . . . . . . . . 105 Other Separation Indices . . . . . . 115 A New Separation Parameter Based on Entropy . . . . . . . . . . . . . . . 118 The Thermodynamic Limit for Separation Energy . . . . . . . . . . . . . . . 126 V. RESULTS AND DISCUSSION . . . . . . . . . . . 128 Algal Growth . . . . . . . . . . . . . . . 128 Harvesting . . . . . . . . . . . . . . . 133 v .. ~.- o.‘ . v.-. -. r ‘ q. TABLE OF CONTENTS - Continued CHAPTER Page V. RESULTS AND DISCUSSION - Continued Sedimentation . . . . . . . . . . . . . . . 133 Centrifugation . . . . . . . . . . . . . . 159 Electroflocculation . . . . . . . . . . . . 163 Filtration . . . . . . . . . . . . . . . . 167 Electrodecantation . . . . . . . . . . . . 171 Other Methods . . . . . . . . . . . . . . . 175 VI. SUMMARY . . . . . . . . . . . . . . . . . . . . 186 VII. CONCLUSIONS . . . . . . . . . . . . . ... . . . 189 VIII. SUGGESTIONS FOR FUTURE WORK . . . . . . . . . . l9l BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . 193 APPENDICES . . . . . . . . . . . . . . . . . . . . . 205 Appendix A - Assorted Observations on the Algae as Cultivated and Harvested . . . . . . 206 Appendix B - Representative Data . . . . . . . . . . 207 vi TABLE 10. 11. 12. 13. l”. 15. 16. LIST OF TABLES Some Examples of Microscopic Algae which have been Cultivated in Cultures Larger than One Hundred Liters . . . . . . . . . . . . . . . Some Grazers and the Algae they Eat . . . . . . Gross Nutritional Characteristics of Some Algae. Experimentation Plan . . . . . . . . . . . . Schultz Medium . . . . . . . . . . . . . . Modified Allen and Arnon Medium . . . . . . . . Zarrouk Medium, Slightly Modified . . . . . Ten-liter Continuous Culture Media . . . . . . Computational Scheme for Deriving the Enrich— ment E as a Function of Dimensionless Time, At 0 O O O O I O O O O O O O O O I O O O A Comparison of Candidate Harvest Parameters for a'Good" Harvest and a "Poor" One . . . . . Productivities of the Light-limited Batch CUltureS O O O O O O O O O O O O O O O 0 O O 0 Recovery Time Constants . . . . . . . . . . . . A Glossary of Symbols used in Graphical Pre- sentation of Data . . . . . . . . . . . . . . Size of the Algae . . . . . . . . . . . . . . . Specific Gravity of the Algae . . . . . . . . . The Effect of Gradient Osmotic Pressure and pH on the Specific Gravity of Oocystis . . . . vii Page 23 28 60 63 61+ 65 68 116 123 131 135 139 150 151 155 LIST OF TABLES - Continued TABLE 17. 18. 19. 20. 21. 22. Page Recovery Time Constants Determined in Semi- continuous Centrifugation . . . . . . . . . . 161 The Acceleration of Settling Caused by Electroflocculation . . . . . . . . . . . 156 Pilterability Assessment of the Several Algae . 158 Electrodecantation . . . . . . . . . . . . . . 17” Electrophoretic Mobility . . . . . . . . . . . 175 Grazing by Starved Daphnia . . . , , , , , , , 180 viii FIGURE 1. 2. 10. ll. 12. 13. 18. 15. 16. 17. 18. LIST OF FIGURES A scheme classifying separation methods . . . Chlorella pyrenoidosa . . . . . . . . . . . . Oocystis polymorpha . . . . . . . . . . Scenedesmus obliquus . . . . . . . . . . . . Coelastrum proboscideum . . . . . . . . . . . Chlamydomonas reinhardtii . . . . . . . . . Anabaena gylindrica . . . . . . . . . . . . Spirulina platensis . . . . . . . . . . . Oocystis polymorpha . . . . . . . . . . . Scenedesmus obliquus in classical multicellu- 1ar form. . . . . . . . . . . . . . . . . . Chlamydomonas reinhardtii with detritus and unidentified—bacteria . . . . . . . . . . Electrodecantation cell, all tubing diSCOD? nected C O O O O O O O O O O O O O O O O 0 Close-up side view of the electrodecantation cell in operation . . . . . . . . . . . . . Schematic View of the electrodecantation cell The electrodeless conductivity cell . . . . . Scheme of the continuous flow electrodeless conductivity cell . . . . . . . . . . . . . Imprecision at the rich—lean interface: an interpretation . . . . . . . . . . . . . Rationale behind the parameter G . . . . . . ix Page 1+1 1+3 1+5 1+7 us 51 53 55 57 57 71 71 73 77 77 112 121 LIST OF FIGURES - Continued FIGURE Page 19. Lines of constant E and constant G . . . . . . 125 20. Batch history of the several algae in the fermentor . . . . . . . . . . . . . . . . . 129 21. Scatter in the data when compared with the exponential decay model . . . . . . . . . . 136 22. H35 as a function of At . . . . . . . . . . . 1H1 23. A two-dimensional display of harvesting success 0 o o o o o o o o o o o o o o o o 1'45 2H. Second of a time sequence . . . . . . . . . . 186 25. Third and last of a time sequence . . . . . . 187 26. Trajectory of am idealized alga which obeys the proposed model for sedimentation . . . 188 27. Spirulina platensis, light microphotograph . 156 28. A one-hour time sequence illustrating the clumping effect in Spirulina . . . . . . . 156 29. Sedimentation enhancement due to elec- troflocculation . . . . . . . . . . . . . 16a 30. Vacuum filtration and electrodecantation of Scenedesmus . . . . . . . . . . . . . 172 max WWWUU 6') Glossary of Symbols area a variable number of standard deviations a slope of a filtration graph, t/V versus V concentration of algae in liquid before harvest (g dry wt./liter) concentration of algae in the lean fraction after separation has begun (g/l) concentration of algae in the rich fraction after separation has begun (g/l) the maximum concentration of algae achievable in a given process dilution rate (liters/liter/hour) hydraulic diameter (defined in Table 1a) enrichment CR/CF flow rate through a continuous culture (liter/hour) (as subscript) mixed state or inlet condition a one dimensional parameter expressing harvesting success acceleration due to gravity (9.81 m/sec2) a normalized parameter related to E, H = SF (Cmax -CR) CR (Cmax -CF) height of interface above bottom of settling vessel; expressed as a queue length purity index constant growth in light-limited culture (liter gram/hour) xi Glossary of Symbols - Continued Boltzmann's constant, 1.38 x 10"23 joules/°K growth constant in hours-1 a constant proportiqnal to illumination level (liter gram/hr. m ) (as subscript) lean fraction Naperian logarithmic operator electrophoretic mobility (u/sec/volt/cm) a population, see subscripts F,L,R,Q; also a number of observations a number of sites available for occupancy the number of objects in the average queue in the fraction R a number of moles, type and location indicated by subscripts pressure (newtons/mz) a probability {event in brackets} (as subscript) indicates participation in a queue recovery; the fraction of algae initially present which is recovered in a harvest (as subscript) rich fraction recycled fraction of harvested material, also a radius a recovery index, two dimensional entropy (joules/0K) time, a fixed value time, a variable retention time xii Glossary of Symbols - Continued < volume (m3 or liters) volume (usually liters) see R,L,F as subscripts work or energy (joules) culture density in grams dry weight per liter (g/l) a fixed concentration concentration of a culture just after inoculation (g/l) the lowest concentration at which light-limited growth takes place (g/l) a fraction of occupied sites a mole fraction the vertical dimension in a settling vessel (m) a constant expressing the effect of input energy on (liters/joule) a small increment instantaneous growth constant in hours-l, may be a function of time cell age (units of time, usually hours) mean cell age cell age at division death rate, 1/A is the mean time between cells leaving a population lO-Bm. usually um lO-Bm 10-18m3 a mole fraction, one dimension of RI Rony's index, extent of separation xiii Glossary of Symbols - Continued fl osmotic pressure expressed in milliOSmols/kg. p bulk density 0 standard deviation o a mole fraction, one dimension of RI w angular velocity (radians per second) xiv -.. v" . Qu,-‘ CHAPTER I INTRODUCTION The overcrowded planet foreseen by demographers is a distinct possibility in the not—too-distant future. Competition for land area between habitation and food production will force a trend toward unconventional agri- culture: exploitation of insects, aquatic animals, bacteria, non-vascular plants, and the like. The cultivation of microscopic algae is considered unconventional agriculture. The algae were extensively studied in connection with the space program of the 1960's when it was thought that algal culture might provide the multiple needs of oxygen regeneration, food supply and waste treatment in a man—alga partially-closed symbiosis. Interest seems to have waned with the news that only for missions of greater than 800 man-days would an algal culture in space have any cost advantage over the brown- bag lunch (Lachance, 1968). The two-species culture is thought to be inher- ently less stable than a culture of many interdependent species. This has not deterred modern agriculture from its trend toward local cultivation of single species. The fail-safe control necessary for prolonged operation of the less stable few-species cultures poses an urgent technical challenge. Exploitation of the algae for food or feed has attracted some research effort. As a protein source, the algae can be managed to yield an impressive quantity of protein per hectare, albeit basically a plant protein with the known deficiencies in amino acids essential to the human diet.l Since unwanted algae are a symptom of water pollu- tion, it seems logical that the cultivation of algae in wastewater might be a useful way to remove the plant nutri- ents before they find their way into an unfertilized lake or stream. Moreover, it is tempting to anticipate a wind- fall in food or feed production at the same time. More will be said of this. In spite of the promising agricultural prospects, the microscopic algae have not been accepted as a crop on a commercial scale in more than a few isolated cases that smack of profligate research endeavors. The reasons for non-acceptance of algal cultivation are nutritional, economic and historic. While the algae promise a high pro- tein yield of a fair quality, the digestibility and avail- ability of the protein is less than desired. There appear to be limits to what the human system can accept; taste and texture problems remain to be solved. For algae as animal feed, these same problems exist. In 1970, a ton of algae cost more to produce than the ton of soybeans it might replace in an animal diet. In several cost projections, the harvesting and drying of the algal product amounts to about seventy-five percent of the total cost of production. Half of this 75% cost is assigned to harvesting, i.e., getting the algae enriched from the typical 1 gram dry matter per liter of culture to the forty grams per liter suitable for a drum dryer. Costs have often been discovered too late. For example, a closed culture is relatively contaminant-proof, but it lacks the evaporative cooling which keeps open culture at a reasonable temperature for good growth. So a cooling cost was found to be a large percentage of production cost in closed culture. The need for agitation, large illumi- nated-surface-to-volume ratio, and efficient carbon dioxide transfer seems to have been overlooked by optimistic cost forecasters. So the meal costs more than the menu indi- cated. In the space program, culture volume was to be minimized. Consequently, the species possessing the maxi- mum of growth rates was sought. For this reason, species of the genus Chlorella seem to have attracted the most attention - in spite of the attendant nutritional and technical problems. There is perhaps a second reason for 0'. ...~ on- -.- -... o.. nu.- — .u‘ s ‘1 u s - n- '5 \ the Chlorella connotation in the word "algae." It is a laboratory weed which can only be avoided by careful manage- ment of the culture in which it is not wanted. I suggest that Chlorella is analogous to crabgrass in many respects. I cannot imagine a sensible farmer cultivating crabgrass, however. The algae, then, have received a bad press coverage not only for the hopes which the Chlorella episode shattered but also for the appearance of algae as a symptom of water pollution. There are about 370 genera and 10,000 species among the green algae alone. If technical and nutritional problems disqualify Chlorella from further consideration as a potential agricultural crOp, there is certainly a good chance that a more promising species can be found. The sacrifice involved in accepting a lower growth rate means only that a larger cultivation area must be planned. It is difficult to take full advantage of a very high growth rate anyway; efficient introduction of light to the culture becomes a serious problem or else the surface- area-to-volume ratio gets out of hand. In this thesis, I will concentrate my efforts on the technical problems of harvesting the algae. I will touch on the associated problems of cultivation and washing the harvested product. Except for an occasional remark, I will avoid the nutritional aspects of algae as food or feed. Hopefully, the wolf will not come to the door, but if he should, perhaps this work will help those who must keep him at bay. 0-.- ' '0 .. _ ‘ d .. ‘ . . .. _ ... u: ‘. w .v . ”tr! ! 17‘) III -. ‘v ‘ l ‘ \ -_-’ s ‘ n g 1 CHAPTER II OBJECTIVES Harvesting the algae deserves investigation since by the present harvesting methods, the cost of harvesting is a large percentage of the total production cost. There may be a thermodynamic limit which prevents improvements in harvesting efficiency. I propose to study the possibilities for increased harvesting efficiency by exploring methods for separating the algae from their culture media and enriching the algal suspension to the point where it can be considered "harvested" - or practically so. Classical liquid solid separation methods will be examined in order to quantify the alleged "difficulty" associated with harvesting. Novel methods will be explored for their potential promise (Figure 1). Using several algal species will provide an interesting variable which seldom appears in published quantitative studies. Ways in which a particular algal species can cooperate or hinder harvest will be emphasized. The species I have selected represent a diversity in characteristics; most have been tried in large scale culture. Indeed a method may evolve for more accurate QOHpmmstnpcoo cowpmasoooam soapMucoEHomm COMHMELOL >coaoo mempouozm mafizeo wowxna whopaso mflxmwoeoco cowumpmem mwmoem mmeam mammamwo GOMumnefiwm mwcmpoxm coH moonpoz cowpmasoooam owpoeonmoppomam \cowpmwoam cofluaeomo< Hmoflwoaon Hmowepomam Hmeeonh boomnsm Hmoflcmnooz HMCOHprH>maw mowaom oovcmamdm mo Hm>OEom meozomz monomemamm mcflseummmao mam;om <--.H mmsta selection of promising species based on harvestability. Cheap and easy harvest is a necessary criterion for accep- tance of algal cultivation into the agricultural setting. CHAPTER III LITERATURE REVIEW The problem of algal harvest is casually classed into the broader engineering category of liquid-solid separation. The conventional techniques which apply for solutions containing low concentrations of suspended solids are the ones one might expect to be successful: centrifu- gation, flotation, filtration, chemical flocculation and sedimentation. There is a voluminous literature devoted to the theoretical and practical aspects of these separation methods. Poole and Doyle (1968) edited a 1.6 Kg. tome entitled Solid-Liquid Separation. It includes 5,181 refer- ences in abstract form. Mathematical models abound for the separation of small discrete particles from liquids by settling and filtration. Chemical flocculation and flo- tation are less well understood. Examples of biological solids in Poole and Doyle are brewers yeast and activated sludge; algae are barely mentioned. Aiba g: 31. (1965) devote a chapter to the separation of microorganisms from culture media; They have selected several models which seem to have worked well in what they term "biochemical engineering". Aiba et 21. 9 10 have also considered energy expenditure in the cultivation operation, i.e. mixing and aeration. Little or no reference to the algae appears in their book, however. It seems that engineers have assumed that the technical problems of harvesting the algae are not unique. Such an assumption may be correct; it may be only that the cost of recovery is too great using expensive techniques in which case less expensive techniques should be sought. Freeman (196”) treats the recovery of cells from fluids in much the same way as do Aiba e: 31. Freeman mentions in addition a few unconventional methods of harvesting — or avoiding harvesting - which caught my interest, viz. dialysis culture, surface culture, biphasic growth and electrokinetic methods. Electrokinetic separa- tion takes advantage of the fact that most growing cells including some algae (Ives, 1959) possess a negative charge. The conventional separation methods depend on a difference in specific gravity between cells and medium for their success or in the case of filtration, on cell size and rigidity. Bier's adaptations (COOper e: 31., 1965) of Pauli's electrodecantation idea (Pauli, 1929) seem to be the most promising of the electrokinetic techniques. The suitable configuration for liquid solid separation involves use of the electric field to prevent the accumulation of charged solids on a filter. Filtration runs (i.e. times between 11 back flushings) are much longer than for filtration without the electric field. The conductivity of the liquid phase economically limits the applicability of the method. Cooper g; 11. filtered a clay suspension and an algae-laden lagoon effluent. They estimated for a water of 1000 umho/cm a cost of 35 Kw hr/1000 gal. for a filter handling 1 gal/hr sq. ft. Latex has been electrokinetically concentrated on an industrial scale for years (Murphy, 1942). Some electro- kinetic pumping of fluid through the filter takes place at the same time as electrodecantation. It is a very uneconomic substitute for the same amount of hydraulic power (Osterle and Farn, 1967). The interesting contributions of Kolin (1953, 1955) appear to me to be inapplicable for separation of cells from their growth media. They involve elegant techniques — elec- trophoretic isoelectric focussing, electromagnetophoretic migration - but are more on the order of laboratory curiosi- ties. Electrophoretic focussing seems unpromising since it takes place in a combined pH and density gradient (Kolin, 1955). It moreover requires an isoelectric point which Ives (1959) failed to find for Chlorella. Electromagneto- kinetic phenomena work even on uncharged particles and bubbles (conductivity of the particles must be unequal to the conductivity of the medium) but much energy is required (Kolin, 1953). The method is suitable for measuring 12 conductivity of cells but the possibility of economic separation of cells from fluid is remote. Eckenfelder and O'Connor (1961) consider the separation of liquid and activated sludge, a mixed culture of floc-forming bacteria. The treatment is crude but quantitative enough to facilitate engineering design as was intended. Emphasis is on gravity settling. Eckenfelder and O'Connor point out that settling of discrete particles in a tank is quite different from the settling of floc where time is required for floc formation. Most work on sedimentation harks back to the classic contribution of Hazen (1904). Hazen showed that in con- tinuous sedimentation the settling tank depth had no effect on the degree of separation achieved; only overflow was significant. Overflow is the bulk flow into a tank divided by the horizontal cross-sectional area of the tank. Fitch (1957) contradicted Hazen and pointed out that for suspensions in which flocculation occurs during settling, the detention period can be of greater significance than the overflow rate. The standard method (_____,1965) for determining settleability of activated sludge involves measuring the downward velocity of the sludge-clear water interface in a one-liter graduated cylinder. The inter- face is usually well—defined and in 30 minutes has typi- cally subsided to about the 250 ml. level. The supernatant;h3 swept virtually clear of particulate matter. 13 Wang (1968) presents classical models for centri- fugation and filtration in cell recovery operations. He writes primarily about yeasts and bacteria and recognizes Species differences. He points out the difficulties encountered in filtration of unicellular organisms directly from their fermentation broths. He also brings up the need for washing of the cells if they are to be food or feed; at each washing operation, another harvesting opera— tion is required. For reasons unknown (or untold), cells that have been washed can be successfully filtered. Wang makes cost estimates for cell recovery. A 5.5 pm yeast can be recovered for 1.5¢ per Kg when the inlet concentra- tion to the recovery operations is ten grams dry weight per liter. Cells are washed once. The investment costs are two thirds of that figure, and are included in it. The cost is inversely proportional to inlet concentration; doubling the washing water volume increases the price by half. Wang suggested that future work be directed towards growth of giant cells and use of polyelectrolytes as flocculating agents. Where polyelectrolytes have been studied (e.g. Bergman gt al., 1958, Kquin and Nebera, 1960), the problem has usually been waste water purification or ore recovery, so the edibility of the polyelectrolyte (or other floccu- lant) is not a problem. Alum and ferric chloride are important inorganic coagulants, now long in use. The .. 14.1.40- .1 u u gnu. r 4 nnv ‘. . .“~.+ .’.o ' " u .u‘_ . . 5..-..1 u‘ .- s.' .v‘ —. 7" u .- ‘H' v... a-v — ._..‘- .._ - 'L,,. a n .. 'u- I- t. I ' 14 chemistry of Fe (III) and Al (III) is considered by Stumm and Morgan (1962). In the pH range between u and 7, the aluminum ion forms a tri- or tetra—positive polynuclear species. Reasonably successful stochastic models have been worked out by Vold (1959, 1960, 1963), Sutherland (1967) and Harris 2: a1. (1966) for the elusive concepts involved in flocculation. The basis for the models is that as particles settle at different rates in a fluid, faster particles overtake slower ones and stick together. The models generally fail to predict sediment volume. Anderson (1956) used this same working hypothesis as did Camp and Stein (19u3). Camp and Stein proposed initiation of velocity gradients by stirring rather than relying on the velocity differential between particles settling at different speeds. I did not pursue the history any further, but the Camp and Stein theory was apparently found in need of only slight modification in a thesis by Ritchie (1955, cited by Poole and Doyle, 1968). Anderson (1957B) showed that settling, flocculating cellulose fibers touched only at their concave surfaces. In a startling short paper, Calleja (1970) gave evidence suggesting that in the yeast Saccharomyces pombe defloccu— lation by heat (”60°C) was reversible. Calcium ions were apparently uninvolved. As an adjunct to focculation, flotation has been used for many materials. Dissolved air coming out of 15 solution forms miniscule bubbles which are trapped in the floc particles. In the 50 - or-so flotation abstracts in Poole and Doyle (1968), flocculation is always mentioned as part of the separation process. The larger the floc particle (with a trapped bubble) the faster it rises (Katz and Wullschleger, 1957). According to Lewis, (1922) filtration theory was found to be in disagreement with industrial experience. The method of Ruth (1933) for constant pressure filtration is used to estimate filterability of various slurries even though industrial filtrations are seldom carried out at constant pressure. A recent opinion (Tiller, 1968) is that each material to be filtered needs to be tested; the best theories still fail to span the range of filterable materials. The subject of filter aids and conditioners is related to the phenomenon of flocculation. The character of the filter cake can be altered to advantage by the addition of polymers which bridge the solid particles. Boucher (1946/1947) suggested a filterability index which was to be measured at constant flow instead of constant pressure. It is more difficult to obtain this index with on-the-shelf laboratory geap than it is to do the more usual constant pressure test. Ives (1960) has composed a moving-front type of computer simulation for making sand filtration predictions from laboratory data. 16 Predation as a means of separating solids from liquid has been considered by Rashevsky (1959) who determined that there was a minimum size particle which a large organism - say, a fish - could afford to go out of its way to eat. Rotifers are said (Schulze, 1966) to play an important role in the removal of turbidity from waste- water. Most of the eligible predators produce by-products other than their own growth. These may constitute con- tamination. Predator growth is a one step food chain and is 90% inefficient in terms of protein conservation. If predators are used merely as traps - that is, they are starved until being fed the solid-to-be-recovered and harvested immediately afterwards - one must then consider the economy of keeping the necessary quantity of starved predators on hand (Schulze's turtles, Schulze, 1966). Fecal pellets may be easier to harvest than the disperse uningested solids. A magnetic method of recovering phagocytic pre- dators (such as leucocytes) was proposed by Cutts (1970). Particulate iron or a-ferric oxide was fed to the cells. After the iron was ingested, the cells were retrieved as they passed near a magnet. Another interesting scheme was proposed by Fulwyler (1965). The slurry passed through an acoustic dFOplet generator which was tuned to make electrically Charged droplets that could hold at most a single cell. 17 Most droplets would be empty in a dilute suspension. Drop- lets containing cells were detected and electrostatically deflected as they fell. The anticipated maximum rate was 1000 droplets per second. Up to this point, I have limited attention to a miniscule sampling of references in the general area of liquid-solid separation. The general literature seldom recognizes the algae as posing any unique technical sepa- ration problems. In Poole and Doyle (1968), for example, only twelve of the 5000 abstracts mention algae; another three or four refer to plankton. On the other hand, pro- ponents and opponents of algal cultivation do recognize the difficulty: Geoghegan (1951, Chlorella vulgaris as food), "Harvesting by ordinary filtration proved difficult as the cells bed down tightly; but centrifuging is quite satis- factory." Groggins (1953, Chlorella as protein source), "Harvesting of large scale culture units containing 1% of algae, dry weight, is a costly procedure, particularly if a super centrifuge is used." Mackenthun 8 Ingram (1967) studied algal removal from lake water for drinking water supply. ”Five algal genera consistently passed through the microstraining fabric. These were the blue—greens Anabaena and Aphanizomenon, the diatoms Cyclotella and Navicula and the green flagellate Phacotus." Borchardt and O'Melia (1961) studied "algae" for waste—water nutrient stripping. They used sand filters 24 inches deep. ”When an algal suspension was filtered, a low percentage of the algae was removed by each increment of the filter bed depth and some of the initial material left the bed continuously, this amount increasing with time." l8 Borchardt (1958) (algal removal of pollutants) claimed that Chlorella, Scenedesmus and Ankistrodesmus were too small for filtering. TrSeveral organiSms seem to respond well when treated as though they were activated sludge." He presented a 5-minute sequence of photographs to show settling speed of a filamentous alga. But he failed to mention the name of the promising species. Bogan (1961) reported that Stigleoclonium (sic) stagnatile, a branching filamentous green alga, flocculated well and settled well. Settling was actually improved by recycling the settled floc as if it were activated sludge. At temperatures above 20°C, however, Chlorella and Scenedesmus tended to predominate the culture. Settling characteristics remained good, however, due to the coagu- lation effect of the insoluble phosphate salts produced at high pH levels. Golueke and Oswald (1965) report the same phenomenon and call it auto-flocculation since the high pH level is achieved at high photosynthetic rates in high light intensities (C02 starvation ). Bogan cal- culates that lime can be applied for one-tenth the cost of using artificial illumination to achieve the required pH of 9.5. Wachs gt a1. (1968, Chlorella and Euglena, deep waste treatment lagoons) recorded an effluent biochemical oxygen demand which was higher than that of the untreated waste water influent unless the algae were removed from the effluent stream. Golueke and Oswald (1965, Chlorella, Scenedesmus, wastewater treatment,by-productu5ed as animal feed) 19 have thoroughly researched the possibilities for harvesting and processing sewage-grown planktonic algae. They ade- quately distinguish between the purposes of wastewater renovation and feed or food production. Different levels of separation treatment are compared which would suit the quality required of the algal by-product. They call the harvest process in which I am interested "initial concen- tration" and work with an enrichment of from 200 mg. dry algae per liter of culture to 20 grams per liter. They studied the following methods: Centrifugation, pH floccu- lation, ion-exchange, chemical flocculation, flotation, micro-straining, passage through an electric field, soni- cation and filtration. Most economically promising of these were centrifugation and flocculation. I will review their work in some detail since my own work is modeled somewhat after theirs. Centrifugation. A continuous centrifuge was used in the field scale experiments. It was a disc type centri— fuge with solids discharge nozzles. The disc angle was a sensitive parameter in the figure of merit - energy con- sumption per ton of algae. Extent of algal removal was determined optically and by decrease in suspended solids. With increasing length of separation run, they noticed an increasing concentration of algae in their enriched harvest Stream (as high as 36.5 g/l after a four hours' run - something short of steady state). Their best figure was A “D” .o-v- n‘.‘. ‘9 . V‘- v., .. v 1 (I! 1 I“ 1:. 20 3200 Kw. hr. per ton of algae, 45° disc angle, 3000 rpm, 280 gallons per minute throughput, 200 mg. dry algae per liter of feed stream. This corresponds to 2300 joules per liter. pH Flocculation. Both Chlorella and Scenedesmus seemed to flocculate and settle best at a pH of around 3. Ives (1959) found, however, that the minimum charge density for Chlorella occurred at pH 7 in deionized water. Chemical Flocculation. The cationic flocculants Purifloc 601 and Purifloc 602 were 95% effective at 3 mg/l. A crude experiment showed no immediate deleterious effects on rats fed massive doses of the Puriflocs. Inorganic flocculants included Ca(0H)2 at 120 mg/l and FeSoq at 40 mg/l in combination enough to raise the pH to 10.6. Filter alum (80% A12(SOQ)3. 18 H20) was also used. 70 mg/l was best in terms of mg. algae per mg. alum used ( about 12 mg/mg). Three minutes mixing at a blade tip velocity of 12 inches per second helped flocculation; settling took 15 minutes. At a conference in 1961, Oswald apparently presented a slide which showed a very puckered horse's mouth just after administration of some alum-flocculated algal feed. Autoflocculation was suggested to be a very promising technique. The natural rise in pH which was noticed to occur on sunny California afternoons, is thought to be responsible. A settling pond 0f 12.7 cm. depth was suggested. I“ 21 Flotation was found unsuccessful; 18 flotation agents were tried. The work of Levin 3: 31. (1962) which was referenced described a method of harvesting algae (Chlamydomonas spp. and Chlorella sp.) by reducing the culture pH to 4, then bubbling air through a Su sparger. Essentially complete recoveries were achieved in 20 minutes with concentration factors ranging from 50 to 200. The richest harvest was 59 g/l - dense enough for direct appli- cation to a drum dryer. The cost was too great primarily due to the cost of pH adjustment. At pH greater than 4, the foam is not stable. Microstraining was generally considered a failure in pilot plant scale trials. Either the cells passed through the screen (aperture not given) or it was predicted they would clog a finer mesh screen. As mentioned before Golueke and Oswald had relative success with filamentous algae but Berry (1961) reported "effective" algal removal for various genera - including Anabaena, a filamentous blue— green alga. Berry used a stainless steel mesh with 23 to 65u aperture. Mackenthun and Ingram (1967) reported that during an Anabaena bloom in a Wisconsin lake, a 35umicro- strainer retained only 16 percent of the algae. Golueke and Oswald's treatment of electrokinetic phenomena is very cursory. The electric field accomplished little or no separation. They did notice, however, that 22 use of aluminum and cooper electrodes resulted in excellent floc formation. Sonic vibration had a dispersing effect on the cells. Filtration was disappointing: usually the algae passed through the filter medium (cloth, paper, teflon cloth 14- 20u). Too much filter aid was required to be economic and still effective. Even cornstarch was rejected as a filter aid since the algal enriched cornstarch is not that much better a feed so that it justifies the process costs. Most of this good engineering work by Golueke and Oswald was done with cultures of sewage-fertilized Scenedesmus and Chlorella. In Golueke's efforts with other species (Golueke, 1961), he seems to have avoided fila- mentous blue green-algae because of the toxins they report— edly produce. Porphyridinum cruentum and Synechocystis sp. were among the other species tried; both are unicellular, the former a red alga, the latter a blue-green alga Porphyridium was harvested by addition of an equal volume of 80%-90% ethanol and stirring until the stringy coagulum could be withdrawn. The high cost of alcohol recovery constitutes the major disadvantage to this method. Mass cultivation of several species has been tried. Success was reported for Spirulina maxima, Chlamydomonas reinhardtii, Scenedesmus obliquus, Chlorella sorokiniana, Scenedesmus acutus, and Scenedesmus quadricauda, among others. Table 1 gives particulars with references. 23 Ill SSS Ne cm 3?? .Hm pm pcoEwHo covampHHm Uoom moosoHpcoo Como H comm m.H moaned mcHHzeHmm AosmHv I ooow mooschcoo ooeo>oo we com I OUonz .am mcHHdnHmm Eonwopcmq cm> mam wcwEwHo I Uoom moonszcoo ammo NE omH I oonmz .Qm mcHHdeHQm umm>emz ooMHn empmum Hmconmooo HHuoemchmp AcanV somomMHmm povooo poo .mdoschcoo ommOHo NE m.mm :m.o chmeom mmcoEOU%EMHno H\m om ow mzoem mmoooHo WmIm oHLaoepoempmc >eumsocH xemo mooHommHHHo Aosmav Hmmcmpm mmaeanucmo soon zoomn ammoao H oooom I seams mHHmeoano mzstHQo .m we com w mozmoHpodmw AQHmHV Homeopm I I moo:CHpcoo ammo H ooomm m.Hm .comuo mSEmmUmcmom NE oom mmesm mopsom Aoanv Homeopm I woom mooschcoo Como H oooom m.m >cm5pmo mSEmmomsoom NE on AmmmHv emcHHz meMHepcoo room mooscwpcoo ommOHo H ooom No.0 .m.: .mm MHHoQOHno woesom vogue: omoaedm mSOSCHpcoo mgdeso mHmom Ammo\wxv moHoomm mcHumm>Lmz no Lopmm ommOHo oHoH» no ammo .memqu ooH cozy Lomemq monsvHso CH oopm>HpHso comm m>mm SOHLB oMMH< UHQOUmOLUHz mo mmHmmem oEomII.H mqm nopmn Como H QOQH I xcmeemo demmoocoom mmconmoo>EMH£o mmoHommHHHm omooHocmeha Il.lh&mmHv ompmum NE : mHnmed> .Hm um emxmz omsmHepcoo poo mooschooo ammo H ooom no.0 HomemH MHHmLOHno ocHomE AmomHv woeo>oo NE m.m .Hm um HHomc< mmJMHepcoo poem noumn ammo H oooH I ocmHmcm Estuomoommnm ownaonu AmmmHv oxoSHoo mcmonLo Ioemumn EsoHepomooo ocm onzmo I .ooom I ommOHo xemo I .m.: Edooooommcoam condom oozwoz mmomedm mdoochcoo whopHso meom Azmowwxv mmHommm MCHpmo>pmm no noumm ommOHo vHoH» no Como UmscflycooII.H mqmHmcmwmocH mpmmu .omco mxmq .mcmEDE pom ooom AbooHv .Hm pm pcoEmHo m EdHooE oz maocx Amemao upsoeo w commemcon .oomm CH owommc mHm HmcoHuHoom AsooHv cospHm: w NPCHI Ucm AoomHv uchm .omm3om on oHeummmoooe wjm Azmeo MEHxME mm.o mm.m eH mcIcm mcHHseHem mnaeoexHom m.m m.m m.mH we mapmscoc .aam moamooocmom o .Qam MHHmQOHzo HmaHoHc26 co czoem mpcmcHEdp won II Hm omtz AHonv pmomm w powwow mHmODHHmL .nom50pm woman .mcHHoom HooNHHHnaomHv oomeHo .opmMp ewupHn memes: mssoHHoo AmcmHo umm.mm sea was ss.o mm.m mHIcH chom mesmmcccccm AHooHv Nuwndq AmomHv ESMQCoccme o mmHopmz mums 0p vow HooNHHHnmomHv mommy oemcm pcmmmoHocs .ooHnoImnmosm mcchoneom AmcmHv .Hm.mm.emc was Hm.o om.m @mImH m.em-mm mHHceoaeo wooedom .mwcoEEoo mpHHHanmomHo .3.o w .3.o w .3.U & .pz ago mo w cHowosm wows mchmq compo»: Amm.o x zv pcmhmam< IoHcpmz Iooemo cHouoem mango .wme< meow mo wOHpmHnopomsmzo HmcoHpHapzz mmoeoII.m mqmHco one mos HHmo mHzp .mocmso xm .moxnmoHMLMPsHm anB pmxHMImnm woo mm: :mEHommm mHLB .mmm Ixchnm HHoo mo Ummsmo woMMHuem cm on >08 ESHHmmmHm comm mo mmmn pmcoonflp one .mHempomn ooHMHpcmoch: ocm msHHspoo nee: AxH.H oeccc .xoome ammo HHpoemcchs mmcoEOU>EmH£oII.HH mmome .eaH TIl .Asema .cocemeeo mcHMOMH mam mpcmamcso nonpo one moHumHem .Esom LMHDHHoOHpHSE HmcemmcHo cH Axe.H oeocc .xooom 2mmo .mssaHHco mesmccmcechI.oH mmDoHe 58 P" iii {In ‘lr sl- u‘c, A p.. 27., ‘A- l.>d‘ ’nx 59 distribute system might result in different recommended concentrations so I will attempt to consider a range of values - say 0.5 to 2 g/l for the culture and 10 g/l to 30 g/l for the harvested material. The harvesting methods I will try include sedimen- tation, semi-continuous centrifugation, filtration using a membrane filter, and variations on these. Each species will be subjected to each method. In addition I will investigate the possibilities for taking advantage of PhYSiological quirks peculiar to certain species such as SWimming, gliding motility, attachment to surfaces, auto- flocculation, and as prey for a suitable predator. Addi- tion Of chemical flocculants will be avoided since I do not intend to perform the necessary nutritional evalua- tions needed to determine if the harvested algae could still be used for food or feed. Table 4 displays the harvest method—species intersections to be studied. In some cases, adequate and convincing research reports can be found in the literature; these will be acknowledged. EXPerimental investigation will proceed to the point Where} an intersection's promise can be estimated. In a few Cases such as sedimentation and filtration, some quan— titative results will be obtained since there is little by Way of numerical comparison in the literature. Some of the intersections are absurd: vegetative Chlorella, for example, is non-motile. Nevertheless, Table 4 presents the gross experimentation plan. .~H. HaH -Su n H «You .sz: Hr~ rUAfiIInHHI I I \s .lVau 1.x 4.x..x. 6O mceeeceesc mammomo< IIIII‘ mHmcmpomm MCHHSLHmm HHuoemncHoe mmcoeoo>EMHno wHomo owHH mo umoe wcHedo mHnHmmom EdmoHomooer uoc assumMHooo oHnHmmoa mnasoezHom poo memmooo oHoho owHH wo pmoe mcHedo mHQHmmom modoHHno you mDEmoomcmom oHnHmmoa mmoowocoemm poo mHHoQOHrB coHumHOHm moHoon ImHSUOOHm memy coHp coHpmm COHpmw wonpoe coHpmooem conoLo< oepomHm HmoHeuomHm Iouonm ImepHHm Iswwepcmo IcoEHUom uwo>smm cMHa cOHpmocceHemameI.e mqmscideum Bohlin. Chlamydomonas reinhardtii is from the lxaboratory of D.T.A. Lamport (MSU) and Anabaena Ellincjrdca Lemm. was given to me by C.P. Wolk (MSU). §RiPUJLina platensis is culture number 1475-4 of the CambrtiHmooocMHHSEHm new asoHHso mxmo nmpHHm mpcm>mpm “OOHMOHOOH thQMHOQ one an3 omHHmadm aposoa OHLHOOHM .HHmO COHHMHCMOOOOLHOOHO one wo 3OH> OHmemzomII.:H mmome 74 wz<¢m£m2 map—.430 .20.; wzJO¢FUme m2<¢m§m2 mF>JOmFO me < + 30583.”. Ea. “woomhomI—w Fm \<\\\\VX\\\\\V\\\ \\\\\\V\\\V\K\\\\\§VT l\\\\\\\\\\\\ \\\y \\\\\\\\\\ x\\\\X\\\\\ I _ i 5.: 0 59:6” uh50m OMAN 2<>zint (1000-1100 psi). surface tension vanishes so the IDInc>perties are strange: J-—j’—C1u.id can boil in a non-violent manner. After 5 minutes the elevated pressure, pressure was slowly decreased 511: The 5-11- the chamber until ambient pressure was reached. 8 an‘lple was removed and kept in a dessicator until the c2":>c‘at.‘ting Operation. In coating, the algae were first C2<:’é3fted.by vacuum sputtering a layer of carbon (0.005um) I?‘ <:’:LlLowed by a layer of gold—palladium (0.005um). The carbon L SE; 1I‘esponsible for the generation Of secondary electrons ‘ 80 which creates the image. The gold prevents local charging by providing a conducting path by which electrons can pass from the sample. The measurement Of cell size was by microscopic Observation Of cells usingéacalibrated eyepiece reticle. Random selection of cells was attempted by a random m twiddling Of the x and y microscope stage positioners and f selecting the cell appearing nearest a zero point on the reticle. Cell specific gravity was determined by a density gradient method (Pertoft, 1967) suggested to me by M. Jost (Michigan State University). A density gradient was estab— lnished in a 10 ml glass centrifuge test tube by subjecting ea 60% (approx.) aqueous dilution of colloidal silica (TIJudox HS, Technical grade, E.I. DuPont, Wilmington, IDelaware) to a gravitational field of 14000 x g to 27000 x E; Ifor 12 minutes (Sorvall centrifuge, SS34 head). The Eilnéidients were then transferred to a swinging bucket centri- fingge (Precision Scientific Co., restored antique). Algal S L1Spension - about one ml - was gently added on top of the Eilfhéiciients. After five minutes at about 200 x g, bands of at:Légéal cells can be Observed in the gradient and compared “7:1-1:}1 colored beads of known specific gravity which are also in 8erted into the gradient (Beads 4 mm in diameter from an easLJ3t3<>nmtive anti-freeze hydrometer, Meijer Thrifty Acres, (2153 :l‘iflbration in a sucrose solution, Handbook of Chemistry 81 and Physics, Chemical Rubber Co., 59th edition). It can be shown that a sphere of constant specific gravity floats in a linear density gradient in a position such that its horizontal great circle is at its specific gravity value in the gradiant. Algal specific gravity was also checked occasionally by drawing the algal band into a tared 100u liter or 200u liter (A) volumetric pipet (capillary con- E striction, S.G.A. CO.). The weight increment was compared with that of the same pipet full Of water. The difference between 1.00 and the cell specific gravity was reproducible in this method to within five percent. The high-sodium <:olloida1 silica gradient had an osmotic pressure of 373 ntilliosmols/kg. near where the algal band was located. (:1 osmOl/kg = 27 atm. approx.) In the case of Spirulina, 23é1rrouk medium was used instead of water to dilute the 1411(10x. Osmotic pressures were estimated by measuring the 1FIneeezing point depression using an Advanced Osmometer (Upicivanced Instruments, Inc., Newton Highlands, Mass.) The 3‘7'53 MOSm/kg osmotic pressure was considerably less than tikléaxt of the Spirulina medium (440 mOSm/kg) in which the C3€3§3.le had been growing; pH was about the same - 10.0. If L“'éa-TZGEr moved due to the osmotic difference between gradient and growth milieu, it would have tended to move into the ( ga~S~vacuole-less) cells, making them less dense. Since 1: ‘A7éiss impressed with how much greater than 1.0 the density VvéEi =3; 3 I accepted this possibility for error. Spirulina 82 was actively motile after 30 minutes in the colloidal silica and appeared healthy in all other microscopically observable aspects. In the case Of the green algae, the osmotic pressure Of the medium was 17.5 m0sm/kg. Ludox HS gradients made with distilled water as diluent average 38 mOSm/kg, a condition that would tend to cause water movement out Of the cells and give specific gravities that were higher than actual. Ludox LS (low sodium) has an osmotic pressure of 26 mOSm/kg and pH 8.5, still unacceptable. I dialyzed 164 g Ludox LS against distilled water. After dialysis 'there were 200 g Of colloidal silica with an osmotic g>ressure of 7.5 mOSm/kg. This was used for gradient work urith the algae other than Spirulina. The gradients are Ineade with four ml dialyzed Ludox LS and three ml H20. C3fllamydomonas was still motile in the gradients after 12 rn-lllll’lutes at 200xg. Other methods to arrive at cell specific gravity C Cleall sinking velocity by slow motion photography (Haddad éiIiACi Lindegren, 1953), or a bulk method (Aiba 33 31., 1964)) I>€3~3.e before the relative simplicity of this one. The balance was manufactured by Mettler Instruments C2<2>171>., Hightstown, N.J. (0 to 160.0000 g 1 0.0001 g). 1: 1“leave relied very much on a gravimetric filtration method. ¥ 83 Procedures Measuring Techniques The method I selected for measurement Of culture density as well as rich and lean stream densities after separation is essentially the same as the suspended matter test for activated sludge (Standard Methods for the Exami- nation Of Water and Wastewater, 12th edition). A filter which will retain the algal material (but preferably not the associated mucilage or bacteria) is oven dried (103°C) and weighed. The filter is then placed in a filter holder and washed with a few milliliters of filtered distilled beater. After a known volumetric quantity of algal sus- pnension has been filtered (usually a vacuum filtration) .aJid the filter cake rinsed with 10-15 ml distilled filtered vaéiterg the filter membrane, paper, or mat, is returned to ‘tldte oven for drying and re-weighing. Culture density (:E3Lispended algal matter) is computed by dividing the dry PVeafllght Of the filter cake by the volume of fluid in which in: tdas suspended. The values of culture density were (illiifte reproducible (for Spirulina, 3.0u filter, standard Ci'EE‘-‘\IJ‘Lation was 1.6 percent of the mean). The selection of tikleat ,proper filter medium was most important. I found that .3 . . . ‘ (j‘udn membrane filters give the most conSistent results. I3 1231: Iless expensive fiberglass mats (Millipore glass pre— f- :L':1-1:eers) were used whenever possible since they are cheaper. (31%; :l‘eilnydomonas and Chlorella penetrated the fiberglass 84 filters, however. In these cases, 0.45u membrane filters were used. Whatman Nos. 1 and 41 papers retained Spirulina but the papers are very hygroscopic and can hardly be weighed accurately as they acquire moisture at about 1 mg. per minute. The dry algal matter might be 25 mg. in such a case. The glass prefilters were relatively non-hygro- scopic as were the membrane filters. A volatile solids (total solids (103°C) minus ash (600°C)) determination on the filter cake wash water indicated that the filter cake retained dissolved solids that were about 40% volatile and 60% nonvolatile. This seas probably mucilage, possibly some bacteria and perhaps tflne debris liberated by osmotic rupture Of some cells. fFlle growth medium for Spirulina is more salty than other ggzrowth media (18 g. total solids per liter). Osmotic Iatlpture of Spirulina cells which had spent twenty minutes jLII distilled water was evidenced by the obvious liberation C>1? phycocyanin (fluoresced red in ultraviolet illumination). F'j'L.ILtrate bacterial dry weight amounted to a calculated (3 ~ (35% of the algal dry weight (5x106 cells per ml, and es timated 1pm radius,coccoid shape, therefore 0.005 g (35:73? weight per liter). The bacteria which stick to the algae are likely the preponderant proportion Of the bac- ‘t:€33rVial population. Ward and M0y8p (1966) estimate that ‘jFJrl ‘their unsterile algal cultures, the bacteria never a JIIInrelated differences Observed in Optical density. For eJ‘CEmele (my data), a culture of Spirulina (0.555 g/l) had an Optical density (425 mm) of 93. If the cells were ES‘:>1'licated for 10 seconds, the Optical density was 68. ‘gx:1‘£§éil cell counting was rejected because of the great CiTjL'1Fiferences anticipated in cell size over the range of Q JIL‘Egéae tested. So by a process Of elimination, filtration 86 was selected as the single method most suitable for measure— ments using the seven algal species. Even so, filter medium changes had to be made to suit the species. And on occasion, a particularly rich and unfilterable sample had to be measured using an optical density method. The filterable suspended solids method was the primary indicator of the density Of the ten-liter cultures Of algae. Management of the Culture The simple mathematical treatment by Pipes and Koutsoyannis (1962) will be followed. Most of my cultures were started as batch cultures .in relatively sterile surroundings and with an inoculum c>f about 100 ml of algal suspension at about 0.5 g dry vqeight/liter. In unlimited batch culture: 28 dt = kX (l) ‘VV11sere X is the culture density in g dry wt/liter t is time - in hours k is a constant specific growth rate (hours-l) But with light being the "limiting nutrient" for (ZVLJZLture density in excess of X8 where shading begins, dX _ V Hf - K for X << Xc (2) '7 here V is the culture volume, Xc is that culture density at which the incoming light energy just compensates via photosynthesis 87 for the respiration losses, and K is the constant growth rate permitted by the constant influx Of light. (l.g/hr). X will generally be much less than XC in my experiments. K is proportional to the illuminated area (A). K 2 LA (3) where L is a constant proportional to the illumination level. In a well-mixed continuous culture algae are removed at the rate X(F/V) where F is the flow rate of medium and harvest in liters per hour. For dilute cultures as dt = kX = X(F/V) = (k - l/tr)X (4) vehere tr is the retention time in hours and is defined as V7/F. If the culture has grown beyond XS where light I>eecomes limiting, dX 3? = K/V - X/tr (5) ~E3<134ation (l)integrates to X = Xoekt (6) where Xi is the concentration inoculated 1E3 0 is required for damping of synchrony, a ratio of o/Ad = 0.5 produces an error Of only ten percent. Thereforeaequation.tl4) will suffice for this discussion. For the uniform age distribution, A = Ad/2. In the light-limited batch culture growth is linear with time. Linear growth may be considered exponential growth with a non-constant specific growth rate. From equations (2) and (7) 93 dX = K and x = 5 t’+ x V s 921 z (15) dt KX (where K18 a variable version Of the specific growth rate k) _ l K - WX (16) K s and 1n2 V Ad ln2(t K X8) (17) In light-limited batch culture, Ad increases with time as the culture grows. For a culture which grows exponentially up to a density of X8 and linearly thereafter, one must consider the transition between growth modes. My data show continuity between the exponential growth region and the linear growth region. Since the onset Of shadowing is probably not a sudden phenomenon there is no. theoretical reason to expect a discontinuity either in the function or its derOvatove Therefore at the transition time ts, S + f(X) 1 I l~+ I I W, 1.1 d A Adj7A The distribution may be temporarily short of newborn cells and a transient dent in the distribution could arise and persist until damped. The Ad boundary moves to the right at the rate (1n2)t while the age groups progress along their cyclic course at rate t, thus some damping is assured. Proceeding to the light-limited continuous culture, the condition maintained in my experiments, the mathematical treatment is similar. Equations (5), (9) and (15) yield K = 1 ,r (19) XV -t ft ,. —%4-e P-t (l - e -t ‘/tr r . _ 1n2 _ XOV -t”/tr -t”/ Ad - K - (1n2) "R“ e + (1 - e tP)tP (20) where t” is the time measured from the start of continuous Operation - at which time (t” = 0), the culture density is XO grams per liter. Since my cultures were batch grown to the desired culture density and the retention time tr was set to maintain that culture density, equation (10) states t l xov r in steady state Operation K 95 Consequently Ad = (ln2)tr for all t” and there is apparently no time lag involved. In sum, Ad is a fixed value (1n2/k) during exponen- tial growth; it increases linearly with time during light- limited growth (ln2(t’ + 3 XS)); and it is constant again (trln2) once continuous Oéeration (light limited) is begun. Meanwhile the mean cell age A follows at Ad/2 plus or minus small, damped transient deviations caused by the changes from exponential to light-limited to continuous light—limited cultivation. My routine for establishing a culture (data from Scenedesmus) was to inoculate at about Xi = 0.005 g/l (t = 0) then to batch-grow to the desired Operating density, say X = 1.0 g/l. Exponential growth would proceed at k = 0.07 hr.l for about 60 hours (equation 6) at which point (X8 = 0.333 g/l, t8 = 60 hrs, t’ = 0) shading limited growth. Thereafter growth followed equation 7. K/V is 02033g/l/hr for Scenedesmus (See Results, Table 11). SO growth to 1.0 g/l took an additional 20.2 hours (t’ = 20.2 or t = 80.2 hrs). Then tr is set at 30.3 hours to maintain the culture density at 1.0 g/l. In the exponential growth stage, Ad = 9.90 hours. In the light-limited batch stage, Ad = 7.0 + 0.693t’ for 0 i t’ < 20.2 96 i.e. Ad is increasing from 7.0 hours to 21.0 hours. In the light-limited continuous culture Ad = 21.0 hours Even though there is no change in A ( or steady state A) d caused by the last transition, no data were collected until at least one retention time (30.3 hours) after the start of continuous operation. This gave any cell age distribution transients fifty hours to be smoothed. It is fairly certain thatIIwould have settled down by this time; my data suggest nothing to the contrary. Differences in physiological characteristics of the algal cells are important when considering the possible application Of algae as a nutrient trap for polluted water. In a properly designed system, the algae will be starved for the nutrient they are supposed to trap. Since light limited growth is probably the best that can be provided when the nutrient is in luxurious supply, the nutrient- starved growth rate will certainly be less than the light limited growth rate with ample nutrient. This is equiva- lent in my cultures to reducing the flow of nutrient solu- tion (F) until the culture density responds to it in other than the hyperbolic manner expressed in equation (11). In general tr becomes very large and the cells become, on the average, very Old. Cell division is prevented due to lack of sufficient nutrient. The effluent solution, its algae removed, is also very low in that nutrient-—as 97 planned. To wait one Of these very long retention times beyond adjustment to the desired culture density would have cost too much research time. Furthermore, Operation in the continuous culture mode would have little point since I was prepared at that point to stop cultivation Of the algae and sacrifice the culture to science. Very dense cultures (old cells) were harvested as if in a batch culture. The major point to be made is the inherent differences in cell age between the fast growing cultures designed for feed production and the slow growing cultures designed for wastewater treatment. A mathematical statement of the dichotomy between cellular productivity and substrate (nutrient) removal is found in Aiba 33 31. (1965) p. 114. Their work is for an exponential growth situation, but the conclusion carries over to our case. The idea Of recycling algal biomass to achieve a high population Of starved cells (desirable for waste- water treatment) has probably never received serious thought due to the reputed difficulty of harvesting the algae. If recycling is considered, a mass balance on a single culture vessel in the case of unlimited growth is _ F F - V(mX) - VX + kX (21) 923 dt where r is the fraction of the harvested algae recycled. 98 In the case of light limited growth dX _ F F K a? - v(I‘X) - v(X) + V (22) In the steady state continuous culture dX/dt = 0. and E a .1. a E l (23) V t V X(l-r) Obviously Unerecycled cells are not getting any younger in the process. The result is that with shorter retention times the desired level Of nutrient removal may be accom- plished. But the cells are still aged. Before leaving this topic it must be mentioned that Nooney (1968) approached the mean cell age concept using a stochastic formulation rather than the deterministic form used here. With the stochastic formulation (in which standard deviation is computed as well as the mean), the standard deviation of the mean age of an exponentially growing population is found to increase with time while the mean value agrees with that found deterministically. The stochastic formulation should be applied to the case of light-limited growth. Procedures Followed in Harvestigg the Algae Sedimentation. This was done in batches of 500 ml of culture. The culture was drawn from the mixed fermentor at a level 5 cm from the bottom of the vessel. It was collected in a 0.5 liter or a 1.0 liter separatory funnel, sampled immediately and let stand at room temperature and 99 illumination. After a measured time interval (when some separation was apparent) the lower cell-rich fraction was carefully withdrawn, its end point being judged by eye. In retrospect, it is here that the greatest scattering of data occurred. Rich and lean fractions were both mixed and sampled using in most cases the filterable suSpended solids method. Total rich and lean volumes were also measured so that a material balance could be determined. In the case Of Spirulina, pre-treatment by brief 5 n/m22 5 eXposure (10 sec.) to high pressure (5.2 x 10 atmospheres) made settling possible. Unless their gas vacuoles were collapsed by pressurizing or sonication (Lehmann and Jost, 1971) the algae sedimented only very slowly. Neither did Spirulina float especially well of its own accord. Flocculation. This is a variation on the sedimen- tation technique whereby the algae are electrochemically flocculated prior to settling. This is accomplished by placing aluminum electrodes into the separatory funnel, then passing through the liquid a direct electric current (500 ma.) for 30, 60 or 90 seconds. Voltage was noted. Settling followed usually for about 60 minutes. Although I agreed not to try additive chemical flocculants, I felt this method deserving Of attention. Alum (A1K(SOL})2 . 12H20) produces an unpalatable product, and the toxicity limits Of the sulfate ion are set at 250 mg/l. The aluminum ion 100 toxicity level is not set, nor does it produce the taste characteristic of alum. Centrifugation. About 5 liters Of algal culture were placed above the semi-continuous centrifugal bowl and sampled. The head was adjusted to about 70 cm. before flow was started. The centrifuge was brought to a constant speed, flow was started and measured. After about a minute when an equilibrium of sorts was established, a sample of the effluent was taken and compared optically (425mu) with the sample of feed stream. Centrifuge power consumption was recorded during flow. Flow was then stopped and the centrifuge brought to a new speed. The hydrostatic head was re-set and the entire process repeated. The centrifuge Speed range was selected to show the region of poor separation. Filtration. Vacuum (10 cm. Hg abs. pressure) was applied under a wetted 30m pore-size membrane filter. A sample of algal culture was then poured on top of the filter at time zero. Filtrate was collected in a graduated cylinder inside the vacuum flask. Filtrate accumulation was measured at ten-second time intervals after starting. Electrodecantation. The apparatus is shown schematically in Figure 14. Electrolyte circulation was started. The filtrate chamber was filled with distilled water just to overflowing into a graduated cylinder. A 500 ml sample Of the culture was drawn. Algal culture 101 was let into its side of the cell. Pressure head was maintained constant by adjusting culture flow just to the point where there was no flow out of the rich-stream exit port. Thus all the exit flow from the cell was clear filtrate. NO measurements were made until the current through the cell stabilized (i.e. the distilled water was replaced by filtrate of higher conductivity). Periodi- cally the rich algal accumulation was flushed out. Voltage was set and maintained constant. Filtrate accumulation was measured as a function Of time. Surface attachment. Tared samples Of glass, poly- ethylene, teflon, polycarbonate, and stainless steel were slowly withdrawn (1 cm/hr) from a stirred beaker of algal culture. The polymers were roughened; one sample of glass was pre-treated by exposure to concentrated HCl; another sample to NaOH; a third was simply allowed to accumulate whatever organic scum it would retain. All glass samples were rinsed in distilled H 0 before the test was begun. 2 Phototaxis. This applies only to Chlamydomonas. Light (10000 lux) was applied from above a one liter graduated cylinder filled with culture. A second cylinder was illuminated from beneath. Samples were taken from tOpemuibottom Of both cylinders after 15, 30 and 60 minutes. Predation. Daphnia magna cultivated by Mrs. B. Burk (Michigan State University) were kept in a starved state. Twenty organisms were placed into bottles 102 containing 20 m1 Of algal culture of known density on day zero. The bottles were kept in the light to provide the crustaceans their necessary oxygen. After 48 hours, the Daphnia were collected by straining the culture through 80-mesh brass cloth. The filterable suspended solids test was run on the algal culture. Controlsvfitmout Daphnia were also monitored for growth. Fresh material from the fermentor was harvested whenever possible. Sometimes in the centrifugation tests, accumulated harvested material was used even though it may have been as old as 12 hours. There was definitely a difference in, for example, filterability Of freshly harvested Chlamydomonas and filterability of the cumula- tive continuous harvest (average age 7.2 hours). The differences suggested that fresh material is more harvest- able than old material. There was no provision for refrigerating the cumulative harvest. Simulation of an industrial process would most likely involve the continu- ous delivery of fresh culture to the harvesting device. Methods 6, 7 and 8 were tested more for success or failure than for the quantitative results sought with the other methods where some degree of success was rela- tively assured. Measurement of Success Since "harvest" has been more or less defined, so also has a quantity which I will call enrichment: 103 E = CR/CF (24) where CR is the concentration of dry algal material in the rich harvest stream, CF is the corresponding concentration in the feed or input stream to the process. Acceptable enrichment will be on the order of 20-40. Since this is bracketed, the criterion for success- ful harvest weighs more heavily on a factor known as recovery, R, the fraction of dry algal material in the feed stream which is recovered in the rich (or harvest) stream. R = VRCR/VFCF (25) where v refers to the volume of the fraction indicated by the subscript. R can be rewritten in terms of concentrations alone. Letting vf = 1 material balances can be written down for algal mass: CRVR+ CLVL: CF (26) and liquid volume: vR + vL = 1 (27) where the subscripts R, L and F refer to rich fraction, lean fraction and feed. Solving simultaneously for v and substituting R in equation (24), cF (CR - cL ) R: (28) 104 In some cases (centrifugation, for example) CL<< CR and R 2 (CF - CL)/CF (29) When the material balance on a process was in order (<5 percent error) the recovery as calculated by equation (25) agreed with recovery as calculated by equation (28). Data were rejected if the materials did not balance within this limit. But computation of the material balance was not possible in all cases: e.g., the rich fraction was not continuously discharged from the centrifuge, and the weight Of algae collected by the predators could not be measured directly. By itself, the recovery in a harvest does not describe the process completely. For example, two possi- bilities for the separation of one liter of culture at 1 g/l dry algal matter are: CR = 7 g/l vR = 0.1 liter )R = 0.70 CL = 0.333 g/l vL = 0.9 liter and CR = 3.5 g/l vR = 0.2 liter )R = 0.70 CL = 0.375 vL = 0.8 liter The first is obviously a more successful process, even though both have the same R. Clearly, without another parameter, say E, information is missing. 105 Prediction Of the Behavior Of R with Time in a Batch SettlingiOperation The plane separating a settled vessel into rich and lean fractions will be the horizontal plan which passes through an algal concentration equal to CF where CL< CFI>l,0), (>>l,l), (1+,l/2) and (l, indeterminate). These last five initial conditions are hard to reconcile with the experimental results and this lends credence to the surviving set (30) which I have selected as reasonable and real. The arrival rate of algal cells on the surface discontinuity that separates rich and lean fractions is stochastic. Interarrival times depend on the population density (CL) above the surface and the terminal sinking 106 velocity. The time course Of settling will also depend on the area presented for settling a given volume of culture. The algal accumulation at the bottom may be thought of as a number of parallel unserviced queues (unserviced if there is no flow out of the bottom). The arrival of cells at the end of the queues occurs at the rate at which cells drop-out (die) from the lean fraction above. The death or drop-out rate (A) is a function of the population size (NL); the probability of a cell dropping out between an arbitrary time t and At later is expressed as Pr {NL(t+At) = NL(t) - 1} = ANL(t)At (31) Growth during the settling phase will be neglected P {N (t+At) N (t) + l} = 0 (32) r L L and changes Of more than one cell will be improbable. Then Pr {NL(t+At) = NL(t)} = l -ANL(t)At (33) The Outcome Of this classic model for simple stochastic death is a binomially distributed temporal population. (Bailey, 1964, p. 91; see also Hillier 8 Lieberman's (1967) limited source queuing model). The probability of there being N cells remaining in the lean fraction at time t is: L Pr {n = NL at time t}: NF! (34) N , (N _N ),(e'Nr*t) (1 - e’*t>Nr‘NL L' F L ' (notation, mine) 107 The probability that the rich fraction contains (NF-NL) = NR cells at time t is the complement: P {n’ = N at time t} = r R N N -N NE! (1 _ e-At F (e-At) F R - ) NR! (IJF-RTRTI (35) The equations imply that complete settling can be realized after an infinite wait. The average cell population decreases exponentially in time, as expected. - . - t NL - NFe (36) while its complement NR increases on a saturation curve . At N = NF(1-e ) (37) R The standard deviation is also time dependent. -At 1/2 -At(l - e )3 (38) oNL or NR = (NFe N N and N can be numerical populations or the number L’ F R of cells per unit volume. In the latter case they are easily transformed into concentrations of dry algal matter per liter, assuming an average cell dimension and specific gravity. To be dangerously specific, I have found that a rich stream density of sixty grams per liter is probably seldom achieved by the inexpensive techniques I have tried, certainly not by settling. A centrifuged pellet might contain 200 g/l but it no longer behaves as a liquid. 108 Dussart (1966) estimates 5.5 uug dry weight for a Chlorella cell Of 2003 volume. A reasonable model for 3.36pm diameter algal cells attaining a maximum density of say 70 g/l is that each cell is in the middle of a 4.3um cube of medium and all other cells are excluded from an occupied cube. Closest packing Of these cubes gives a Cmax Of 70 grams Of dry algae per liter, there being 12.6 x 1012 cubes per liter. SO a culture density Of c = 1 g/l has 180 x 109 occupied cubes per liter or a pOpulation per unit volume (N) of 180 x 109 cells per liter. Since N is proportional to C, -At CL = CFe (39) OI‘ -At (40) L F Another fact can be gleaned from the mathematical analysis. The standard deviation on the average number Of cells which have left the lean fraction applies as well to the same cells as they arrive at the end of close- packed unserviced queues at the bottom of the vessel. The number of queues is the bottom area divided by the area of a 4.3um cube or 49 x 109 queues per m2, a large number. In a vessel containing separated fractions, the boundary moves upward as more algae settle. The reduction in volume of the lean phase was ignored in equation (31)ff, it being usually a very small percentage Of v the volume L, 109 of the lean fraction. It is, however, a considerable factor when considering v which is small early in the R process. In fact the interface defines VR' The separation level has been set somewhat arbitrarily as that level at which the algal concentration is C It is a plane which F. cuts across the ends Of the queues of close packed cells 9 . such that (CF/Cmax) x (49 x 10 )A cubes are occupied. Since C < Cmax there will be some lean phase also in the F plane filling CL/Cmax of the cubes. If the separation plane intersects the queues h cubes from the bottom, then the difference between C /C and C /C must be made F max L max up by queues that are longer than or equal to h units long. There are (C - C )/C x (49 x 109)A Of them. F L max That is to say, h is located where the probability of a particular queue being longer than or equal to h is (CF - CL)/Cmax At 2h.}= (CF - CL)/Cmax = (CF/Cmax) (l - e ) (41) r RQ in a queue where N is the number of close packed cubes RQ in a single queue. The probability Of there being h or more units in the queue at time t is the same as the probability of there being fewer than N -h units in the lean phase F directly above (and destined to arrive at) a particular queue. Modifying equation (35): 110 h N I N N -N pp {N >h }= I FQ, (l_e-At) FQ(e-At) FQ RQ RQ - Iq =0 NFQI(NFQ-NRQ)I R0 (42) where the subscripts FQ, R0 and LO refer the population N to the number Of cells in a single queue or its special extension into the lean region. This expression is difficult to evaluate for large values Of NFQ' The binomial probability distribution for large -At values of N (e ) (1 - ext) is approximately the same as FQ the normal distribution (Papoulis, 1965) with mean N = N (1 - e_At) and standard deviation RQ FQ At 1/2 0 = (N e'*t(1 — e‘ )) (43) FQ Figure 17 is a visual display Of the situation for large At. The question to be answered is, how fuzzy is the boundary between rich and lean streams? More importantly, is the quantity Of lean fraction which is separated along with the rich fraction going to reduce C by dilution Rmax and impart to it a time dependence? The area under the normal curve for values Of N >h represents the proba- RQ bility that a queue is longer than h. Over a large number of queues, assumed equal at t = w, this is the percentage of queues longer than h and is, of course, computed by the error function. If 60 percent of the queues are longer than say 0.5h, then the concentration measured at a plane cutting through the vessel at this 111 level be 0.6 Cmax + 0.4 CL. The first fraction usually predominates. The cumulative distribution function, then, gives a pictorial representation of the boundary between the two fractions (Figure 17). Since it represents C, and the numbers in the queues represent depth, the area under the curve represents total quantity of each component contained beneath a boundary, say h: h h A f E dy and C = fiL—QQX (”Hau5) C R fh d O F o Y where C/CF is measured from the Pr = 0 axis and y is the vertical dimension in a settling vessel of constant cross sectional area A. The area under the cumulative distribution func- tion Of the normal curve is computed numerically (in Abramowitz and Stegun, 1964). The area further than one standard deviation from the mean constitutes only 4.6 percent Of the total area outward from the mean itself; beyond 20 it is 0.085 percent. Little error will be introduced by assuming that lean fraction penetrates below h as far as the mean, and the entire volume below the mean is occupied by packed cells at a density Cm . ax - I > ' For (h NRQ) lo, the error in CR When At, the dimensionless time, is small, the can be disregarded. normal approximation to the binomial distribution cannot be used. This is logical since when the mean queue length is small, the symmetric normal distribution would 112 .coprvmsasmucH c< "mommnmch CMOHIQOHQ one pm conHoosaEHII.sH mmome 113 20.0mm z0 (46) Pr { IN Since this will only be applied in cases where ao>NRQ (i.e. close to the bottom of the vessel and early in time) the case NRQ<[NRQ-oo) has zero probability and the only way NRQ can be outside the interval is to be greater than (NRQ + as). If h = NRQ +86, (47) then Pr {NRQ :(NRQ-+ao)}=Fk~{NRQILh} (48) a’2 1 CF/Cmax (l - e‘At) (49) From equations (37), (43), (47) and (49), h and ac can be numerically determined as functions Of time. Corresponding values of CR are computed as above, assuming, as before, that having defined h, the level of mean queue length in the rich domain does a pretty good job of equally dividing the volumes Of packed cells and intruding lean phase, i.e. CR = (NRQCmax + ao CL)/(NRQ + a0) (50) Use Of the Tchebysheff theorem will cause CR to be conservatively low in the region where it is used. The 115 normal distribution can probably be trusted for NRQ 3 o. In Table 9 are tabulated some computational results. The model is applied to a one-liter settling vessel 40 cm2 in horizontal area, 25 cm deep. CF = 1 g/l and NFQ = 832. C is 70 g/l. max Other Separation Indices Separation indices other than R and E include an interesting one by Rony (1968). It is called the extent of separation and was designed to describe chromatographic separations. ll 12 e: abs det for binary separations (51) Y21 Y22 where Yij is the number of moles of phase i in region j divided by the number of moles Of i initially present in the system and 6 is the "extent" index. Its alleged universality is appealing but for typical algal or sludge separations it can be simplified. Grams substance per liter is analogous to the molar quantities prescribed. Consequently if material 1 is water and region 1 is the lean region, (1000 - CLlVL 2 1 for most algal harvests. Y : 11 (1000 - CF)VP (52) 116 TABLE 9.--Computation Scheme for Deriving the Enrichment E as a Function Of Dimensionless Time, At. Cubes 4.290 on a side cmax = 70, cF = 1 cells 3.36udia, 20 3vol. NFQ = 58275, 832 will be full at CF 2 1 At 0 /c (l-e-At) N o a h E = c /c F M RQ R F .0005 .00000714 .416 .64 (374 (240 >l.l2 Tcheby .001 .0000143 .831 .91 (264 <24l >1.24 Tcheby .002 .0000285 1.66 1.29 4.02 6.85 17.7 Normal .005 .0000713 4.15 2.03 3.80 11.86 25.1 n .01 .000142 8.28 2.86 3.63 18.66 31.6 " .02 .000283 16.5 4.02 3.45 30.3 38.5 H .05 .000697 40.6 6.22 3.20 60.5 47.3 " .1 .00136 79.2 8.46 3.00 104.6 53.2 " .2 .00259 151 11.1 2.79 181.8 58.2 " .5 .00562 328 14.1 2.54 363 63.2 " 1 .00903 526 13.9 2.36 559 65.9 " 2 .0124 720 9.99 2.24 742 67.9 " 5 .0142 827 2.36 2.19 832 69.6 " lO ' .0143 832 .195 2.19 832 70 " 117 Similarly Y = ELXL = (l - R) (53) 12 C v F F Y21 = (1000 - C3)VR 5 R (sq) : C V - Y22 CR R - R (55) FVF Then 1 l-R e: abs det (56) R/E R will be close enough. Some more manipulation yields _ l-R 2 1 -€-(m)(1 - (R/E) ) OiRil (57) In all but the worst harvesting situations, E> 5. 8° 1 ' 5‘ %—3—§7E (58) Substituting for R and B using equations (28) and (51), one finds that 1 - R _ CL 1 - R/E ’ a; (59) In working with CL/CF’ we are using something close enough to e to be assured of its generality. DeClerk and Cloete (1971) deduce a relationship for the intermediate local entropy level and derive from it an approximate local quantity called the purity index of the separated region. Ij 3 - loglonij (60) 11.. 33 118 where nij is the number of moles of component i in the jth separated fraction and Ij is the purity index defined for the region 0i<211 <:O.1. 33 In our case the concentration of algae in the lean fraction is certainly small enough so that Ij would be defined for the lean fraction. The fraction of water in the rich stream, however, is greater than 10%, so the index would not be meaningful. Said (1964) had the right idea in keeping the recovery index as a complex number. RI(v,¢) where RI represents the recovery indes and (v,¢) are respectively the fractional impurity and size of the fraction. moles of impurity in a fraction (61) moles of main component ¢ : moles of main component recovered. moles of main component at the start of the process (62) A New Separation Parameter Based on Entropy, Using the entropy of a completely separated culture as reference, the ratio of the total entropy of a par- tially harvested system to the entropy of a perfectly harvested system is a new parameter which has some inter- esting properties. I dub it G. Its definition is based on the entropy-of—mixing conceptvflfixfllcan in turn be 119 derived from information theory (see Past (1962), p. 129, for example). The validity of my parameter depends on Shannon's proof (Shannon, 1964) that entropy is additive. (I am indebted to Prof. M. Krzywoblocki, Michigan State University, for raising this point.) In brief, the entropy of a mixture of Nox particles of type A with No(1-x) particles of type B is related to the number of ways (m) in which the particles can be distributed over No sites. (No is the number of sites in a given volume and is also equal to the total number of particles; x is the fraction which is type A). No! m : INOX)I(N°(1-x))! (63) Appealing to the cornerstone of information theory, 8 = kBln m (6a) where S is the entropy above a reference level and k is the Boltzmann constant, 1.38 x 10-23j/K° Using Stirling's approximation. In N0! = Noln No - No (65) we can derive: s = - NokB{x ln x + (l-x) ln (l-x)} (66) Thinking of the particles as the occupied or unoccupied cubes, x is analagous to the volume fraction of occupied cubes, viz. 120 C C and No is now the number of cubes max per liter. (67) X: Figure 18 depicts the derivation of G: G 2(3 + sR)/sF (68) L G is a measure of the completeness of mixing, the comple- ment to the completeness of harvesting. It is normalized on <0,l> Table 10 presents a comparison of several parameters which might be used to measure the success of a harvest operation. It can be argued that CR and CL are really all that is required and their orthogonality is guaranteed. On the other hand they are dimensioned quantities and neither is normalized. R is well known in the literature but E is not. Neither is E normalized. A normalized version of B must involve the maximum value that E can attain. Let H be the normalized version of 1/E: Cmax H _ l E:max-13 : E§(Cmax-CR) (69) Cmax ‘ E - C C —C ( max ) R[ max F) The choice of Cm is guided by the data. The case for ax keeping a two dimensional vector is strong; of the three pairs shown, (CL/CF’ H70 malized. Furthermore it will be found to behave well in ) is the only one which is nor- the model described. Notice (Table 10) that if the parameters are taken singly, they can all (except G) be made to misrepresent the quality of the harvest; only the double combinations and G survive the test. 121 FIGURE 18.--Rationale behind the parameter G. 122 '11::352751 {3 PERFECT --‘3:-'-' HARVEST’ 121 FIGURE 18.--Rationale behind the parameter G. 122 PERFECT HARVEST’ 123 .H\m on on 0p cmep mw xme o 0 0 00 0 H 0 a 0 a a +0 0 a 000>000 02 0000. 0000.0 0 00.0 000.0 000.0 0 00.0 00. 000.0 00. 0 0 000 0300.0 00.0 0: 0.0 00.0 000.0 0.0 0.0 0.0 0.0 0.0 3.0 0 000 000.0 0000. 0.00 000.0 00.0 0000. 00.0 000. 00:. 0:00. 000. 00.0 0 000 0000.0 0000. 00.0 00.0 0000.0 0000. 00 00.0 00. 0000. 00. 00 0 0000 000.0 00.0 0 0. 000.0 000.0 0 0.0 0.0 000.0 0.0 0 0 0000 000.0 0 0 a 0 0 00 0 000. 0 0:00. 00 100000 00 0000000 0 0 0; 00 .,000 00\00 rrm 0 0> 00 0> 00 00 xcom . 0C0 :rHOOnm : .m ocm pmo>gmm :Uoow: m pom mbmwm8m9mm pmw>bmm opmoflocwo mo comflhmmaoo p0mcmo m on: mopsyaso Ham 00:9 Om pwflam @809 0 hp omymdmpm opoz Mpmo .aowcmEhmm may :0 mmmam Hopo>om may mo >90pm0£ £0pmm||.om mmame 130 .0052: 02.0 0.0 00 0.0 00 00 o... 00 0.0 E 9.2:... ._ . m0\ \\ E33280 u \o 0 0:93.20 AInOZV acoon°c< k \m\ 028:9: 0_.§ooo\0n \ £500 0 \ O \ 5 005502625 0 \ o unsaouocoom\ u :90 .-o.N . rim (I/5)x 131 TABLE ll.--Productivities of the Light-limited Batch Cultures Slope K/V Estimated Alga g/l-hr. Xs (g/l) Chlorella .01585 0.35 Oocystis .0326 0.35 Scenedesmus .0337 0.32 Coelastrum .0160 0.38 Chlamydomonas .0278 0.30 Anabaena (N03’) .01565 <0.2? Anabaena (N2) .00304 <0.3* Spirulina .0178 0.25 X5 is the culture density (under my conditions) above which the culture is light limited. Slopes are taken at 0.6 g/l. The slope K/V is also the productivity attained in steady state in continuous culture (Equation 11). *Insufficent data. 132 seems to hold over the other algae, especially the green algae, may stem from the fact that the Schultz medium on which all the green algae were grown was designed for Scenedesmus cultivation. I have no explanation for the decreased slopes that some of the cultures exhibit late in growth. Possibilities include burnt out fluorescent lamps which always went out three-at-a-time (one eighth of the light), a change in pH which went unnoticed, a change in viscosity which might have affected the degree of turbu- lence (and exposure of individual cells to the light) or production of a growth-inhibiting substance by the algae themselves (Pratt 33 alia multa, 1944). In the Spirulina cultures, one must be careful not to exceed the CO2 demand of the algae by too much; at the high pH of the medium (pH~10); enough C02 can be scrubbed from the atmosphere so that no enrichment is necessary. But even the air flow rate must be kept low until the algal biomass has grown up - otherwise a precipitate appears in the culture (CaCO3?). The constant growth rate and productivity of light limited cultures means theoretically that several species can be grown in the same culture and one will not necessarily take over. Of course, this reasoning assumes no antagonistic behavior beyond the over-growth threat. As evidence for mutual growth, I found that my nitrate- grown Anabaena culture had a constant level of Chlorella 133 contamination for a week of continuous (not batch) oper- ation. (1 g/l algal biomass, Chlorella estimated at 20%, pH 7.6, l/8x Allen and Arnon Medium). Chlorella's reputation for overrunning other cultures it contaminates has probably been built on experience with batch cultures; it may tolerate a high culture density (shading?) better than other algae, but in continuous culture, the culture density can be kept low enough to offset any advantage Chlorella might find in a high culture density. Harvesting The various harvesting methods will be considered as topics; the algal species are a parameter. Sedimentation (As concluded earlier, two variables are needed to describe fully a harvest's relative success. Having selected CL/CF and H as suitable, the data were processed into that form and plotted on graphs. (The hypothesized model which was presented earlier was derived only after consideration of these experimental results. Hence these data, being the basis for the model, cannot be said to validate the model, rather the theoretical approach was taken to instill confidence in the trends deduced from the scattered data.) CL/CF versus time (not shown) confirmed the suspected exponential decay of the population in the lean fraction. The exponential time constants A for the 134 algal species and subpopulations were computed using a least squares linear regression analysis (on log-transformed data) which suppressed the constant term and ensured passage of the curve through the point (0,1). Ais set in tabular form (Table 12) along with a predicted value of CL/CF one hour after sedimentation has begun. Sub- populations were identified by studying a graph like Figure 21, "picking off the wild ones" and checking back to see if for some (any) reason they could be grouped. Identifi- cation was "confirmed" by ascertaining statistically significant difference in their mean time constants (A) at the ninety percent confidence level. In this way a behavioral difference between young and old cultures of both Oocystis and Chlamydomonas was noted. This is an elaborate method for checking out a "hunch" based on the experience with the algae.' In truth, one cannot at this point differentiate between cell age and culture density. It may well be that CL/CF = f(CF) in which case the model is incorrect and CL/CF is not the well behaved parameter (independent of CR) that is desired. Other experimental evidence will bear on this point, generally in favor of cell age as the distin- guishing feature. A quick appraisal of the tabulated values of A indicates that treated Spirulina and Coelastrum are the fastest settlers while young Chlamydomonas and Anabaena 135 .3000 >0n00moo0mm50 00 mcflappmm mammnmc< ApmeMHHOO mmHOdom> mmmv 0000. 00 000.0 .000. 0000.- 00 0000.- 00000.- 000000000 0000. 00 000.0.000. 0000.- 00 00:00.- 00:000.- 00000000000 000000000 0000. 00 000.0.000. 0000.- 00 0:000.- 00000.- 00000 0000500050000 0000. 00 000.0.000. 00000.-nr.0:0000.- 00000.- 0000000 0000500050000 0:00. 00 000.0.000. 0:00.- 00 0:00.- 0:00.- 5000000000 0000. 00 030.0.000. 00000.- 00 00000.- 000000.- 00500000000 0000. 00 000.0.000. 0000.- 00 00000.- 00000.- 00000 0000.000 0000. 00 000.0.000. 00000.- 00 00000.- 000000.- 0000000 00000000 0000. 00 000.0.000. 0000.- 00 00000.- 000000.- 000000000 000mmchmLMQ £0 0 :o 0m>00000 0:0060 HM>000C0 mocmoflm oo womv mocmoflmcoo woo HI . .00500u0 00 o\0o .mdepmcoo 0508 >0m>oomm-I.00 m0m36 hours), 7.7 versus 6.15 for young Coelastrum (I=10.u hours). This change in C/N may be the phenomenon responsible for increased specific gravity with increased mean cell age (or whatever). 150 .00005 £0300m 050000005 50 0000000> 00050 000 0005m00 020 0000 059 550 5500 00 00005005 0000 055m 00 00005005 05050000 .500.0-0.0 00 00005000 00050000 00 005555 050 00 z 550 500000>05 50055000 050 0000 .0500 0050o00nmm 0 00 000005005. .0 00 00000 000 .50w500 550 500000 0000 500 0500500mm .55m0m .>0U .500 550 Enhmm 5005 5003 50050000005 000000505omx0 0000050xo0mam .m5o0 550 00 55m 5000 0m500 005050000 0500005< . MCOHHMKVQmwQO 005ow0500o 00050 050 000 o 050 .0 .0 00053 m\0Aonmv 00 0050005 00005005 000500005 050 m0 ma 000 5.00 m.m5 00000000 #000 0.0500 womv 000005 0500: 0 000 0.0 0.00 00000000 00000 0000 m500000m 000.0 555000 >00>00m 0000o0am 000 o.m m.mm 0000500 5500000000 000 0.0 0.00 0\0 0.0 0.0000 0000 000000 0000: 0 0000000000 00 0.00 0.00 0\0 00.0 5000000000 mm m.m m0.m 0\m om.o 0050505>50050 00000055 00 0500050500 050000000 O0 00005000 000050000 00050000000050 50 0005 00505005000: mm 00.0 0\w 0.0 m550000500m :000: 0000000 00 0.0 0.00 0\0 0.0 00000000 00>00 000 00.0 00.0 0\0 00.0 0050000500 000 00 050000005 000 m0500> 5005 050 0 000 00.0 00.0 000 00.0 000000000 2 5:0 55:9 500005500 0050050 00w0< 050 mo 0N0mnl.00 mqmfiB 151 0005 0 05000000 00000 500.00v 0.0 :0 m 0 Aomo.0Immo.0v 0.00 mm.o A00500 5 00005000 I I I I hmo.0v 000w0v00000 00050050v I I I I 0000.0Iomo.0v 00000 0.0 mm m0I m.o0 o Aomo.0Imuo.0v 0.0 0.0 mm m0I. «.00 0.0 nmo.0Immo.0 03v m.m 5500000000 50 m0I «.00 m , 000.0 00.0 I I I I 000.0 0.00 00.0 05500005000 5 m.m0 m 0 A000.0Imo0.0v 0.0: 0.0 m.0 :0I m 0 Aom0.0Imo0.0v 0.50 00.0 m.o 05.00 m Amz0.0Im0.0v mmo.w 0.0 m0 0 5 0000.0Immo.0v 05.0 000. 00000000 000000055 000.0Iomo.0v 0.0 m0 0 5 0.0000 000.0Immo.0v 0.00 0.0 A00000055 mmo.0Io>o.0v I I I I 0.0000 oo0.0-omo.0v 0.00 0000 .0. o.m. m.m m 0.0 Ammo.0Imbo.0v 0.00 o.0I 000000000 .000m .005 .000M .005 >00>00w 00000000 000500V 0m0 0\m x 5 5 «mm «mm 00000W005 h000500 0050050 00000 050 00 500>000 00000050II.00 00000 .mmndp map mo Eouuon may um oazm mm swan mm mcflcc59 cam wommgdm nmma: may pm bra gum: pcmflvmam mm m Oman ma mpmce .pcmwvmpm mpwmcmc may cw vcmn Hmmam map ummc cmhdmmme ma mafi 152 m.m m.ma m m omo.a mammnmc< :pcmommcmm: I I I I oma.a mm mmb.a I I I I ooa.a ma.m mmu.o I I I I Am3H.HIHmH.HV om :mo.a mSEmmUmcmom A.uaso pmxmnmv mum mm: Ha Ha Ammo.anmo.Hv nupmn H\m a Aku Asepmnv IHLSmmmgmv mum mm: Ha Ha Ammo.HIaoo.Hv .an :m mjb. mafiaspflam ©.oH :.:H m m.© Aozo.HImmo.Hv mm.© mm.o omo.a h.ma mm. m.ma m.:a w w m:o.H b.ma m. mmcoeovmemano .vmnw .vme .vmpm .UmE >pw>mam owmwomam Ampsonv mmm H\m x = n «mm «:a dame wmme zuflmcmw wLSpaso < UmSCHPCOOII.mH Mdm7.l <10 1.076 1.076 1.067 :lu 1.086 >14 1.091 1.098 Oocystis was cultivated at pH 7.1, 14 mOSm/Kg. collapse most ( 90%) of the gas vacuoles (Figure 27). I used tank C02 and ran the static pressure up to about 5 atmospheres, gauge. A dramatic color change occurred at this pressure which could not be produced at a lower pressure. For sonication, 22 watt-seconds was sufficient to treat 100 ml at 2.21 g/l dry Spirulina - or 220 joules/ liter. Coelastrum behaves very well as a settler as long as there is no scouring (turbulence near the rich-lean interface). This alga can be resuspended with very little energy expenditure per unit volume. An important consid— eration in cultivating algae which settle well is the agitation power required to prevent unwanted harvesting in the growth chamber. At an agitator power input of about 1.1 watts per liter, for example, a Coelastrum culture- 156 .oommmaaoo mmaosom> mow "pmmq mmaosom> mom QPHS "pnwflm "coax .SQMLMOPO£Q0hoflE pgmfla .mflmcmpmHm MGflHSLflamII.mN mmwam .HocozmonpflcfloI:.m SE m omoom comp mm: pmoa one so nmxmmn opp OH .mceassflam :fl pommmm mafiaesao one wCflPMmeSHHH mocmsqmm oEHw ado: mco 9m>oommII.nH mqm mpw wo m.o ow c300 om U\ 0 wow ow oesao> was: 9mm pdacfl mmpmcm« m A .mwuscws om CM mcflappom Amflmmn 0\ UV pcmam>flnvm muscHE om m m>mficom ow oESHo> was: 9mm pdmcw mmsmcm+ Mpmo pcoon0mSmcH mcflasgflmm .comwpmmeoo pom mammn o: “mappmm you 0H0 pan oopmnmamm mammnmc< maaoo manor omom comma Hmaooo. mmcoeoozemano .Ucoammp #0: who Ednpmmawoo .o.fl .mcflaupmm HmEpoc 0cm mcflauuow oopmasoooam comzpon moocopommwo oz Enhpmmamoo coma comm ammooo. magmmemcmom maamo nae oossm comma momoooo. mflpmsooo comma OQNOH msmoooo.+ maamgoano om > so > Am\avs mpcoEEoo mm 50w .1 x.l . H\ a om +A3V wom.fizv COHpmHsooononpomHm mo ommsmo mcflappmm mo COflpmpmHmoo< mQHII.mH mqm 1 g/l 266 Scenedesmus 928 Coelastrum .5 g/l 12.6 Coelastrum > 1 g/l u.0 Chlamydomonas ~.5 g/l 190000 Chlamydomonas > 1 g/l 310000 Chlamydomonas .58 g/l 23800 ("Palmella") Anabaena N03 (y) .5 g/l 55600 Anabaena N03 1 g/l 29H00 (old) All tests were performed at 76 mm Hg absolute pressure in the vacuum flask, Bum pore size membrane filter. *Single measurements only. The differences among the species are the significant factor to note. 169 through - a matter of five minutes on the small filters I used. Coelastrum was the opposite; filter cake an inch thick was not uncommon. The difference in filtration specific resistance between the motile swimming Chlamydomonas cells and the sedentary Palmella—stage cells is quite dramatic. A mixture of the two cell types filters quite rapidly for a few seconds but then the filter clogs quite abruptly. Undoubtedly an interesting mathematical analysis could be undertaken on the subject of filtration of two species, one a high resistance species, the other a low resistance species. The reader is spared such an analysis in favor of the common sense hypothesis that in the mixed population, the number of filter clogging motile cells did not accumulate enough to clog the filter for the first ten seconds. Proof that the Palmella stage was indeed a stage and not a contaminant species was under- taken. A single 5-cell clone of Palmella type cells was withdrawn from very dilute suspension using a homemade micromanipulator. This was then cultivated in a shake culture and produced a good population of motile cells. I also observed (once) a Palmelloid clone break up into four motile cells which then swam away. The reddish eyespot characteristic of Chlamydomonas was also visible in the non—motile form. Note here the couple of instances in which old cells filtered more easily than young cells (Chlamydomonas and Anabaena). Note also that they are 170 the two worst filterers. The more usual case is repre- sented by Chlorella, Oocystis and Coelastrum and I submit that old cells filter more easily than young cells. By way of comparison of numbers, Trubnick and Mueller (1958, referenced in Eckenfelder and O'Connor, 1961) reported 26% x 109 cm/g (units changed, r adjusted for compressibility) for digested and activated conditioned bacterial sludge from a wastewater treatment plant. If that is any indication of an economic limit for vacuum filtration, it means that only Coelastrum and possibly Oocystis (old) would be eligible for this method. Here again the range represented by the algae is impressive and it certainly makes a case for unseating Chlorella as the typical alga. The energy requirements for filtration are nominal. At constant pressure (the manner in which my tests were conducted). deV de = P = Energy per unit volume The pressure just happens to equal numerically the energy required per unit volume. This comes to 0.9 x 105 n/m2 (or nm/m3) W/V = 90 joules/liter and is independent of the filter cake resistance. The pump inefficiency is another matter. If maximum power is being transferred, then the maximum power theorem puts the pump dissipation 171 at another 90 j/l. So the job gets done for around 200 j/l exclusive of backwashings and investment. Electrodecantation The shape of the filter clogging curve is shown in Figure 30. Electrodecantation data are plotted to the same scale. The straight line continued past 70 minutes before it was found necessary to flush out the enriched algae from the collection chamber. Longer filtration runs is the advantage claimed for this system. Chlamydomonas and Chlorella turned yellow-brown in the chamber - quite possibly because of the excessive heating. Current densities of 300 ma/cm2 at field strength 33 v/cm give a power density of 10 watts/cms. Although the principle of operation is supposedly removal of the negatively charged algal cells from the filter membrane surface, this is only part of what occurs. Figure 11 shows the algal harvest slumped down (gravity) and some little bit attached to the electrode chamber membrane. Filter cake formation is effectively prevented. The extent of migration to this last mentioned membrane is evidence for the electro- phoretic effect expected. Movement of filtrate through the filter and filter cake is driven by the electric field. Electrokinetic pumping is the major effect: if the electric field is shut off, flow virtually stops (even with the 0.15 m hydrostatic head forcing fluid through the filter). Reversed polarity causes reversed 172 .Umm: mnmz mhmpafim wcmanmE Eno.m .Hijm zampmsflxogamm ohm mmmam pmpaflm .Aa\w o.Hv m26mmoocmom Mo coauMpcmowponpoon 0cm soapmppaflw Endom>II.om mmame Amy—52...: w}: no 0.... ow 0,» cm 0.. o . . 3 a 4 q . H. m . .o. N as . Iou m zofifizgmoomhomd m \ w 0 3 Ion ) . M 292,5: :39; ( +0¢ 173 pumping. I.e., after operation for one half hour, the apparatus is essentially electrokinetically pumping fluid through the clogged filter. Table 20 presents the results of my experience with this apparatus. The dismal energy consumption was difficult to accept. Electrolyte was changed to Na SO so that 0 would be evolved (rather 2 4 2 than, say Cl from NaCl) in the hopes of recovering energy. 2 The amounts of H2 and 02 generated are small; the recover- able energy from 9650 joules spent amounts to only 214 joules. In both the filtration and electrodecantation process, recovery is complete (i.e. CL/CF = 0). E (and H) varies depending on the frequency and extent of backwashing. Cooper 33 El. (1965) worked out a relationship for the energy requirements of electrokinetic filtration/ pumping. They conclude that electrical conductivity of the water is the single most important factor in determining power requirements. As an example they predict an energy requirement of 10000 joules per liter for a water having a conductivity of 300 umho/cm in a unit with a flow rate of 0.07 ml/min per square centimenter of filter area. The figure is directly proportional to conductivity and flow rate. They claim, moreover, that this is the upper limit for an economically interesting water purification scheme (2l¢/1000 ga1., U.S. 1972). .Eo\OLE:000 u owco Eo\>\oom\Enm.H n: nmEo\cWE\HE 50.0 n 0 nom Uopmsm0<«« .mu 00\mo mane .ESEMQHE m on ow venom mm: oponpaflm onp wo wmm psono mo oESHo> o .Hooo maponoamo 00o oomam ogp aoox 0cm p30 cowpoopm 0onownao opp nodam OH: .m.a mm: .ma magma no magmaoflymmsa msam> spflaflnoe ++ .ooaanooH sanmnopm .esfieme Npasgom HH<+ 174 .nopoz mop aw oowao voucommdmon w cosmos co mxhoz .nmflz cove mafiasnwmm ++0mbh 0000:m omsm 00.H 0.: A0020 ocooooc< 000:0 ooomma + 00.0 0:.0 omom 00050: + mm.o aa.m mchEOesemaso + mawmoo oou whopaflm .ocow won Esnpmoaoou 00:0: ooomba 000 m0.a mm.a 0 I 0 + m0.a H>.0 00:0: 0000bm + m0.a mm.a mSEmowoaoom o o osoa Hm.m om.o mflpmsooo oomsm ooommm ©3QH Hm.m mm.m mflpmsooo own: 00:maa 0m:H 5mm. mm.a maaonoazo oosaa «0.0 000m00 0m:a smb. ma.m oaaonoanu «avopmsmoo CHE\HE Eooppm woom .Eo\o:E: Aa\mv 00 cwE\aE Lopwa\moH30m opos Loflp sopwa\moadom >uw>flu COMHochoocoo opon Ioswcoo ooom opoguawm .Eo\>,mmI waowm oflppooam mmEo m.b mono Lopawm coflympcmooooppoonII.om mamov Upowcoym AEo\>\oom\mv mm o opsuado mpwmcoc opspaso z Hmmam may mo xuw>flposvcoo HMOHQHHU 0H0 spflaflnoz unpmnozaogpomHmII.Hm mqmwuoucowopaom m xHszmm< 208 a.aa oe Hea.o H.m~ ae.~ = e : m.aa om mNN.o w.w~ me.~ apnea 5.~ Nauaawa m.wa om Hm~.o a.~a mam.o : = e o.ea oe eaa.o e.ma aae.o e e = a.aa ma amN.o k.aa 5am.o emo.o mem.o Nauaeea «.moH oe ~H~.o m.- e .. .. .. . a.aoa om oe~.o m.5~ : e e e Aemuam>uea N.aa ma mam.o m.a~ o~a.o amo.o Nma.o ~5Ha<- Inamamv EsuumoHooo e.ea om Naa.o e.a Nae.o : e e 5.ooa om mme.o a.ma ~a5.o : = e e.aa oe mme.o a.ma aa5.o : z = Auee>pee e.ma ma Nak.o aa.a aaa.o kao.o aaw.o Nepaeam eao-aee0 enuumofiooo o.ea oeN ama.o ew.wa emm.o mmo.o No.a Napaea n.5a oe mem.o e.ea e e e : e.ea om aNm.o me.e aae.o mmo.o o~e.o Napaee a.eoH oeN mam.o ee.e mam.o : e e m.OOa oma mmm.o e~.m mam.o : e e o.~oa oe eam.o e5.~ Noe.o Nmo.o mam.o Napezam m00080005ofino ¢.®© 00 CONeo GomH .. 2 2 2 w.ma om eam.o e.ma em.a : e : N.ooH ma ema.o H.NH am.a apnea em.a Nasazmu m.ma oe eam.o m.mH oea.o : e e e.ma om Hem.o H.ea «ma.o e e = a.ea ma aae.o m.oa oma.o mao.o eNa.o Naaezea Ae.uaoev a A maamHOHLQ Axvmeceaea A.cae005au AH\wvao Aa\w0 o Aa\wv mo AHIH:VQ Aa\wvx memo Hoauoumz wcfifiuuom Avonowucoov cowumuCoEwwom 209 .mHoccsm 0p0uouo0om osu aw oHuuom u.:000 unn .0o055H0 "ocomnos< m.aoH an wea.o aa.oH : e.eo~ em aao.o m.ma 000.0 o.aa oe mea.o e.wa : c.00a om Hea.o a.ma mam.o e.ma ma om~.o m.~a ma.a e.ma om eNN.o 0.x mm.a 5.ma em mom.o N.aa : ~.ma ma eaa.o o.ma nae.o m.m0a ON Nam.o ~.w mm.a Aemueea:0e I oNH mma.o ~.~N m~.H \emuauaaaapav mcwanuwam a.wa oma ome.o k.ea New.o e.aoH oe Nam.o me.ma Nem.o w.ooH om Nam.o Ne.e 000.0 o.m0a men eem.o mo.o mm.a Aeeuemppaav can eem.o aa.e mN.H H5oma~H eeaaapaam Aevmuemaea I.eaevmeaa aa\aoae aa\acae Aa\aoae Aa-peva Aa\avx mama Howuoumz wcwfiuuom AcoSCwucoov cowuoucoEHwom 210 Filterability 30Mar72; Chlamydomonas at 0.712 g/l; filter pore 3 microns. vacuum 125 mmHg abs. Time(seconds) Filtrate volume T/V (m1) 0 0 0 10 5 2.0 20 6 3.33 30 6.5 4.62 40 6.75 5.93 50 6.9 7.25 60 7.0 8.57 90 7.75 11. 120 8.0 15.0 180 8.75 20.57 240 9.25 25.95 300 9.75 30.77 600 11.25 53.33 900 12.50 72.0 1200 13.50 88.89 1500 14.25 105.3 1800 15.0 120.0 The points from 10 seconds to 180 seconds give a straight line of SIOpe 7.90. The correlation coefficient is 0.9 4. Filterability or Specific resistance is 1.97 x 101 cm/g. Specific resistance r = 49.8 x 1010 x slope x pressure drop culture density where pressure drop is in inches Hg vacuum and culture density is in grams dry algae per liter. For comparison, 23Apr72; Coelastrum at 0.95 g/l; filter pore 3 microns. vacuum 125 mm Hg abs. Time(seconds) Filtrate volume T/V 0 0 0 10 140 0.0173 20 203 0.0985 30 254 0.118 40 296 0.135 50 335 0.149 Replicate: O O O 10 149 0.0671 20 220 0.0909 30 275 0.109 40 323 0.1238 50 354 0.141 211 Centrifugation Date CF(g/1) Acceleration CL(g/1) Energy(j/1) Chlorella 12May72 0.563 2786 x g 0 20,000 ” " 1935 0.036 15,400 " " 1238 0 14,100 " " 697 0.165 9,370 " " 310 0.204 7,780 " " 77. 0.521 5,570 25May72 1.853 3792 0.141 27,100 ” ” 2786 0.154 19,300 " " 1935 0.206 15,100 ” " 1567 0.245 13,100 ” " 1238 0.245 11,500 " " " 0.157 12,400 " " 697 0.420 9,350 " " 310 0.664 7,790 Spirulina 15Dec7l 1.06 77. 0.782 9,750 (pressurized) ” ” 309 0.267 10,600 " " 445 0.186 11,300 " " 606 0.270 12,900 " " 697 0.016 12,200 ” " 792 0.032 12,800 19Dec71 1.125 309 0.340 9,540 ” " 309 0 37,688 " " 309 0.190 37,700 " " 309 0.578 11,400 " " 484 0.245 10,400 " " 697 0.215 11,900 " " 948 0.260 14,100 " " 1238 0 16,900 Oocystis 23Mar72 2.44 948 0.0717 8,970 " " 697 0.832 7,450 " " 948 0.752 8,680 " " 1088 0.007 9,340 " " 1238 0.076 9,230 " " 1238 0 11,800 " " 1238 0.059 - Centrifugation (continued) Date Oocystis (cont'd.) 3Mar72 H H H H H H llMar72 H H H H H H H Chlamydomonas 2Apr72 H CF 2 212 (8/1) .635 H Acceleration 1935 1567 1238 948 697 484 310 948 697 484 310 174 310 484 697 2786 H 3792 4954 6269 1935 1238 697 310 CL(g/1) 0.127 0.163 0.173 0.958 0.04 0.04 0.067 0.125 0.375 0.330 0.083 0.059 0.152 0.173 0.105 0.058 0.017 0.039 0.067 0.298 0.491 Energy(j/1) 13,750 11,450 10,125 8,380 7,980 9,170 8673 8800 9072 8851 9412 9063 8964 7920 16,770 16,250 21,250 26,750 33,220 13,060 9,103 7,703 6,534 213 00: 00 :::.0 «0.0H 0:0.0 H00.0 0:0.0 m5oo00H ocHHsnwam .oHuuom no: moov 0:0 HHos mouofinooofim mcoonoo< exec cc use.a ma.ee mca.~ e e e caca cc ca~.H ca.ce ccc.c aecec ca.~ Nauezcw MOOH ONH .qamoo M¢Vom 2 2 2 2 chN can ccc.c ca.c ccc.c amc.c ccc.c Nauezma exec cNH ccc.a ac.c e e e e 5mmm cNH cec.a a~.ca e e e e Nccm can ccc.c NN.~ Nac.a eec.c Ncc.a Naccakm cauasooc c cc cca.c cc.ca Hem.a : : maamzmm ccca cc ccc.c «c.ca ac5.H e e a eeaa cc ccc.c cc.ca ~85.a : a = cam cc Ncc.c 5e.c~ ccc.a eac.c ccc.a «asezem accc cc cac.c ha.c cc5.c e = : caaa cc ccc.c ccc.a ccc.c e e : Neau cc acc.c cc.e acc.c HNc.c ccc.c Neaezca enacaoacc eNcc cc cca.c cc.c cac.c e c.c ~aea<5 cacc cc ccc.c ca.c ccc.c : e e caec cc ccc.c ce.e ccc.c e cac.c Nauacm ccea cc cae.c ca.ca ecc.a emac.c c.a Napacc mama cc ac~.c cc.c ccc.c eacc.c cam.c Napezcm amngocaEeHcc ceca cm ama.c kc.c ccc.c coaec c.~ Nauacca Rama ca cma.c ccc.k Hac.c amc.c ccc.c Naaacam aaacaeacoc ccca cm Nac.a c.c~ cac.a e e e c852 cc ccc.c a.~c ccc.c : e e COMM mm @OmoH Momm 2 2 2 2 Ncca ca cec.a c.~e caa.a kmc.c cac.a Neemcca ecc cc ccc.c cc.m~ ecc.a : = e cmc cc cck.c ce.a~ ecm.a e e = Naca cc Nce.c cc.ae ecm.a e ecm.a Naemcka ccc ca Nac.c Na. N ecawc ccc.c ckc.c Neaecca caeacccecoc Aa\ac A.eaec.Eaa aa\acac Aa\ac a Aa\cc c Aa-paca Aa\acx up.a kwuocm wcwfiuuom cowuofinoooHMIouuoon "11 11111711111111