CONTINUOUS LACTIC ACID FERMENTATTON OF WHEY T0 PRODUCE A FEED SUPPLEMENT HTGH IN CRUDE PROTEIN Thesis for the Degree of M. S. MICHIGAN STATE UNWERSITY ALBERT KENT KELLER 1974 IIIIIIIIIIIIIIIIIIIIIIIIIIIIII ABSTRACT CONTINUOUS LACTIC ACID FERMENTATION OF WHEY TO PRODUCE A FEED SUPPLEMENT HIGH .IN CRUDE PROTEIN by Albert Kent Keller Continuous lactic acid fermentation was shown to be a simple and ef— ficient process for converting whey into a source of crude protein (N x 6.25) for ruminants. Lactic acid bacteria were used to convert whey lactose to lactic acid, which in turn was neutralized with anhydrous ammonia. The process was operated non—aseptically in a lh—liter fermentor for U2 days without degeneration of the product and with an actual in- crease in efficiency of conversion. The effects on the continuous fermentation 0n retention time, number of fermentor stages and pH were investigated. An increase in retention time up to 15 hr resulted in an increase in lactose conversion, but only marginal improvement was realized by further increasing the retention time. With a retention time of 15 hr and a pH of 5.5 in a single—stage fermentor, the residual lactose concentration was 0.7%. Increasing the retention time to 31 hr resulted in a residual lactose concentration of 0.6%. However, by employing two fermentors in series with a total reten— tion time of 31 hr, it was possible to reduce the residual lactose to less than 0.1%. Increasing the pH from 5.5 to 5.8 resulted in a substan— tial reduction in the residual lactose concentration, but further in- creasing the pH to 6.0 resulted in only a small additional reduction. A published mathematical model for lactic acid fermentation was modi- fied and used to simulate the whey fermentation. A close fit was obtained between the simulated and the experimental results. Use of the model in- dicated that lactic acid is produced not only as a function of bacterial growth but also of maintenance metabolism. The simulation predicted that a three-stage lactic acid fermentation would give little improvement over a two—stage process. After 1h days of the continuous fermentation, a significant increase began to take place in the ability of the culture to ferment lactose to lac- tic acid. After 19 days, the increased fermentation rate became stabilized. The change in the culture made it possible to increase the throughput rate by a factor of 3 and yet retain the same degree of lactose conversion. The predominant fermentation product was shown to be lactic acid. Gas chromatography was used to detect other metabolic products, but only traces (<0.2%) of ethanol and acetic acid were observed. The fermented product contained approximately 9 times as much crude protein as the un— fermented whey. Batch fermentations were used to demonstrate that sources of growth factors can be used to reduce the fermentation time. The addition of yeast extract (0.20%) or cornsteep liquor (0.25%) reduced the fermentation time to one-half of that experienced with unsupplemented whey. Batch fermentations were also used to demonstrate that product inhi- bition rather than substrate exhaustion accounts for the progressive de- cline in fermentation rate observed in batch lactic acid fermentations. Both ammonium lactate and calcium lactate inhibited the lactic acid fer— mentation, but the former was about twice as inhibitory as the latter. The interaction of pH and product concentration was evaluated utilizing a model relating pH to the biological activities of weak acids and bases. CONTINUOUS LACTIC ACID FERMENTATION OF WHEY TO PRODUCE A FEED SUPPLEMENT HIGH IN CRUDE PROTEIN Albert Kent Keller A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Microbiology and Public Health 197A DEDICATION To Carole for her love, patience and understanding throughout this investigation. ii ACKNOWLEDGMENTS I wish to express my gratitude to Dr. Philipp Gerhardt for the oppor— tunity, for his assistance and for his exemplification of persistence in the pursuit of excellence. I also appreciate the expert counsel of Dr. H. E. Henderson of the Department of Animal Husbandry and Dr. C. A. Reddy of the Department of Microbiology and Public Health. Dr. George Coulman of the Department of Chemical Engineering programmed the mathematical model for execution on the digital computer, and 1 sin- cerely appreciate his enthusiastic support. This investigation was supported by the Michigan State University Office of Research Development and by the Michigan Agricultural Experiment Station. iii TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES INTRODUCTION HISTORICAL REVIEW h.l h.2 h.3 Feed Supplement from Whey Lactic Acid Fermentation Centinuous Fermentation MATERIALS AND METHODS \DU‘IWU‘IWU’TWW oo—qmm £4,010!“ Bacterial Culture Substra te Fermentation Fauipment Procedure for Continuous Fermentation Fermentor Measurements Lactose Determination Ammonium Ion Determination Lactic Acid Determination THEORY RESULTS 7.1 computer Simulation of'Batch Fermentation 7.2 Simulated and Experimental Continuous Fermentation 7.3 Adaptation of’the Continuous Culture 7.h Effect of'pH in the Range of'5.5 to 6.0 on Continuous Fermentation 7.5 Quality of'Product from Continuous Fermentor 7.6 Effect of Yeast Extract and Cornsteep Liquor on Batch Fermentation 7.7 Ammonium Lactate and Calcium Lactate Inhibition of Batch Fermentation iv Page vi vii 10. TABLE OF CONTENTS (continued) DISCUSSION 8.1 Effect of Retention Time and Staging on Conversion of Lactose Adaptation of the Continuous Culture Effect of pH in the Range of 5.5 to 6.0 on Continuous Fermentation Quality of Product from Continuous Fermentor Effect of Yeast Extract and Cornsteep Liquor on Batch Fermentation Ammonium lactate and Calcium Lactate Inhibition of Batch Fermentation CD (DO) @CI) 0\ \J'l-lr' UJR) APPENDIX BIBLIOGRAPHY Page 51 51 55 61 62 67 69 1. LIST OF TABLES Table Title Page 1 Values used in the simulation of'batch fermentation, 25 from Luedeking ( 3 7 ) 2 Adaptation of bacterial culture as indicated by a 35 change in residual lactose concentration during pro- longed continuous fermentation at two levels of pH 3 Effect of pH in the range of 5.5 to 6.0 as indicated 37 by residual lactose concentration 14 Material balance data for a batch fermentation of 1&6 fresh whey supplemented with 0.2% yeast extract 5 Values for pH obtained during batch fermentations of 67 whey supplemented with various levels of calcium lactate and inoculated with L. bulgaricus #2217 6 Values for pH obtained during batch fermentations of 67 whey supplemented with various levels of calcium lactate and inoculated with a mixed culture of lactic acid bacteria 7 Values for pH obtained during batch fermentations of 68 whey supplemented with various levels of ammonium lactate and inoculated with a mixed culture of lactic acid bacteria 8 Values for pH obtained during batch fermentations of 68 whey supplemented with various levels of ammonium lactate and inoculated with L. bulgaricus #2217 vi Figure 10 2. LIST OF FIGURES Title Schematic diagram of level-control device. Symbols: 1, stainless steel tubing; 2, rubber coupling; 3, balloon; 4, agitator. Centinuous stirred tank fermentor Effect of'product concentration on the specific growth rate, with data from Luedehing (37). symbols: X, pH 5.6; 0, pH 5.4; a, pH 5.2;A, pH 4.8. Comparison of curves from the computer simulated data with the experimental data of Luedeking (37) far bac- terial density (X) and lactate concentration (0). Effect of total retention time in single- and multi- stage fermentations as predicted by simulation. Symbols: 0, one stage; X, tzvo stages; A, three stages. Effect of’the total retention time and staging on the level of'residual lactose in continuous fermen- tation of'reconstituted whey. The results shown in Fig. 5 were used to interpolate the curve fbr the single-stage (0) between 0 and 7 hr and fbr the dauble-stage (X) between 12 and 31 hr. Graphs of the linear approximation used to simulate the efiect of'product concentration on the specific growth rate with calculated data points indicated (X). Effect of the total retention time and staging on the level of residual lactose in whey as predicted by computer simulation (curves). Emperimental data points fer single—stage (0) and double-stage (X) fer- mentations are included far comparative purposes. Gas chromatogram of'4.4% lactic acid standard with addi- tion of 0.2% ethanol and 0.1% acetic acid (attenuation = 16). symbols: 1, backgroundk 2, ethanol; 3, acetic acidk 4, lactic acid. Gas chromatograms of'product from (top) and feed to (bottom) continuous fermentor (attenuation = 8). symbols: 1, backgroundk 2, ethanol; 3, acetic acidk 4, lactic acidk 5, unknown material. vii Page 12 18 26 28 31 33 3h hO LIST OF FIGURES (continued) Figure Title ll l2 13 1h Effect of three concentrations of yeast extract on the rate of batch fermentation of fresh whey, measured indirectly by ammonium ion concentration. Symbols: :3, 0.1%; o, 0.2%; x, 0.4%; ., typical unsupp lemented fermentation. Effect of three concentrations of yeast extract on the rate of batch fermentation of fresh whey as measured by appearance of lactic acid (closed symbol) and the disappearance of lactose (open symbols). Effect of three concentrations of cornsteep liquor on the rate of batch fermentation of fresh whey. Symbols: A, 0.25%; O, 0.50%; x, 1.00%. Effect of ammonium and calciwn lactate on the terminal pH of batch fermentations with L. bulgaricus #2217 or a mixed culture of lactic acid bacteria. Symbols: 0, culture #2217 with ammonium lactate; as, culture #2217 with calcium lactate; A, mixed culture with ammo- nium lactate;.a, mixed culture with calcium lactate. viii Page 143 A): 147 50 3. INTRODUCTION The American dairy industry annually discards 22 billion pounds of whey on fields or in municipal sewers or markets the whey as products which return marginal or negative profits (2, 71). About half of the milk solids remains in the whey after separation from the cheese curd. With cheese production increasing and with greater public awareness of environmental pollution, cheesemakers are under increased pressure to find alternative means of salvaging this enormous supply of nutritious milk solids. The magnitude of the whey problem is not as great in Michigan as elsewhere, since Michigan produces only 3% of the nation's cheese. How- ever, for the individual cheese producer within the State, the problem is acute. Pressure is also upon processing plants for potatoes, sugar beets, cherries, vegetable products and other carbohydrate foods to find economi- cal means of managing their wastes. A process for coping with whey is potentially applicable to these products as well. A potential solution exists in the fermentative conversion of whey into a feed supplement containing a high concentration of crude protein (N x 6.25). There is a broad base of fundamental technology and several specific processes for such a fermentation (see the historical section below). These processes use lactic acid bacteria to convert the whey lactose into lactic acid, which in turn is neutralized with ammonia. The fermented product contains approximately 8 times more crude protein than whey. After condensation, the product contains about 50% crude protein and serves about as well as soybean meal in the rations of ruminants. The above process has not been developed and commercialized in the past due in part to the relatively low price of alternate protein supple- ments, notably soybean meal, and to the limited information on the perfor— mance of ammonium lactate in the rations of ruminants. In the past year, however, dramatic changes took place in the price of all protein supple- ments. The commodities market, reflecting the world food shortage and particularly the protein shortage, reacted with higher prices for all protein supplements. It is now economically feasible to consider alter- nate sources of crude protein, such as ammoniated whey, which in the past were uncompetitive. In addition, the safety and efficacy of ammoniated whey as a source of crude protein is now demonstrated. The problem of whey disposal and the improved market for protein supplements served as the impetus for research to improve the process for fermenting whey to produce a source of crude protein for cattle. The primary objective was to determine the feasibility of a continuous fermen— tation process, which potentially is much more efficient than a batch process. During the study, the following factors of continuous fermenta- tion were evaluated: retention time in fermentor, number of fermentor stages, pH of fermentation, culture stability under non—aseptic conditions, and product quality. In addition to the laboratory experimentation, a mathematical model of the continuous fermentation was developed and used to simulate the process on a digital computer. In ancillary studies with batch fermenta— tion, stimulation was demonstrated by the addition of crude sources of growth factors; and inhibition was demonstrated by the addition of the product, ammonium lactate. A. HISTORICAL REVIEW h.l Feed Supplement from Whey Whey is the liquid fraction that results when fat and casein are removed from.whole milk in the process of cheesemaking. Whey contains roughly half of the solids and most of the vitamins and minerals of the whole milk from which it is derived (67, 68). Ten pounds of milk yield about nine pounds of whey and one pound of cheese. Whey contains about 93.1% water, 0.3% fat, h.9% lactose, 0.9% protein, 0.6% ash and 0.2% lactic acid or about 7% total solids (A1). 0f the l.h0 billion pounds of whey solids available in the U.S., approximately 700 million pounds are processed and used as by-products. About 250 million pounds of solids are marketed as edible products for human consumption (70), while the balance (approximately h50 million pounds) is marketed as animal feed (67, 68). The edible products are used in baked goods, ice cream, sherbet, cake mixes, batter mixes and the like (68), whereas milk replacers for calves and pigs are the biggest mar- kets for feed-grade whey (67). The lack of suitable markets limits the quantity of whey sold for human consumption, and the return to the whey processor from feed-grade whey is barely enough to cover the cost of pro— cessing. The animal industry could use all of the whey solids produced in the U.S., but only if whey prices were below production costs (68). Pricing below costs is an unsound business practice which few whey pro- cessors can afford to follow. Consequently, alternative methods must be developed for utilizing the 1.15 billion pounds of whey solids which are an economic burden to the cheese manufacturer and an ecological burden for the public. As early as 19h5 it was recognized that whey and similar materials could be converted to sources of crude protein for ruminant animals. Two Dutch patents (27, 50) were issued for batch fermentation processes in which the traditional method of converting sugar-bearing products to lactic acid was followed, with the exception that ammonia was used as the neutralizing agent. A culture of a homofermentative lactic acid bacteri- um, such as Lactobacillus bulgaricus, was used to convert the whey lactose to lactic acid. During the fermentation, the lactic acid produced was neutralized intermittently with ammonia so that the pH in the fermentor remained between 5.0 and 5.8. The fermented product, condensed to 10% of its original volume, was reported to be a good protein supplement for ruminants. In 1958 a U.S. patent was granted for essentially an identical fer- mentation process (10). In the same year a more detailed description of a method for manufacturing a high-nitrogen, low-lactose product from whey was'published in the scientific literature by Arnott, et al. (3). This process was virtually the same as the two Dutch and the American patents. An important observation was that the fermentation time could be consider- ably reduced by the use of automatic pH control. About 80% of the crude protein in the product, referred to as "ammoniated whey", was in the form of inorganic ammonium ions and the remaining 20% was true protein in the form of whey proteins and bacterial cells. Feeding trials were conducted utilizing "ammoniated whey" as a pro- tein supplement for cattle (21). For most of the cattle, feed consumption was reduced when the supplement was included in the concentrate portion of the ration. However, those cattle which maintained high consumption rates performed very well. The conclusion of the study was that there may be a palatability problem.with the supplement; however, the supplement that was consumed performed very well. In a subsequent study, in which "ammoniated whey" was included in the total ration rather than in the concentrate portion of the ration, no palatability problem was found with the supplement. In fact, the ammoniated whey appeared to per- form.as well as soybean meal when used as the sole source of supplemental nitrogen (A5). "Ammoniated whey" produced by Moore, et al., (Abstr. in J. Anim. Sci. 3hz36l-362, 1972) at Auburn University was used as a source of crude protein for lambs. The study indicated that the product was more effec- tive as a protein supplement than urea but less effective than cottonseed meal. Subsequent studies by Alston, et al., (Abstr. in J. Anim. Sci. 36:208, 1973) demonstrated that "ammoniated whey" could not be dried by conventional methods. At the same time that the work described above was being carried out, pure ammonium salts of organic acids were also being evaluated as feed supplements for cattle by Allen, Henderson and Bergen at Michigan State University (1). The results of these studies encouraged Henderson to embark on a program of converting agricultural wastes to ammonium lac— tate. Henderson also demonstrated that a product analyzing 50% crude protein could be produced by fermentation of whey without sterilizing or pasteurizing the system. In 1973 the project was extended cooperatively by the Department of Animal Husbandry and the Department of Microbiology and Public Health in- cluding the design and construction of a pilot plant. Approximately 500,000 lb of whey were processed by Henderson and Reddy to produce about 25 tons of condensed product for use in extensive beef cattle feed- ing trials (23). These feeding trials involved about 150 steers, and the experimental design was developed in cooperation with the Food and Drug Administration to facilitate that agency's evaluation of the whey product as a feed additive. The fermented whey product performed nearly as well as soybean meal in the feeding trials when both were used as protein supplements (22). Alternate substrates are being evaluated by Reddy to expand the potential applications of the technology. h.2 Lactic Acid Fermentation Much of the fermentation technology of lactic acid production is directly applicable to the production of "ammoniated whey". Lactic acid was commercially produced by fermentation as early as 1881 (53), but in recent years the fermentation process has been partially displaced by one of direct chemical synthesis (A8). Because of the previous importance of fermentation to the production of lactic acid, there are a considerable number of process descriptions (9, A8, 53, 56, 73), a review article (1A) and a bibliography (13). Commercial fermentation processes for the production of lactic acid utilize homofermentative lactic acid bacteria, which are capable of con- verting sugar to lactic acid with yields in excess of 95% (M8, 53). Thermophilic, aciduric lactobacilli are particularly useful because they thrive at temperatures (AS-50 C) which cannot be tolerated by most poten— tial contaminants and because they tolerate high concentrations of acid in the medium. Lactobacillus delbrueckii is the organism of choice for fermenting sucrose- and dextrose-containing materials such as molasses and corn dextrose. L. bulgaricus is used to ferment lactose—containing substrates such as cheese whey. Lactic acid bacteria synthesize very few of the vitamins, amino acids and unidentified growth factors required for their growth. In fact, these organisms are so fastidious that they are often used in biological 7 assays for various growth factors (15). In commercial processes utilizing lactic acid bacteria, it is often necessary to supplement the medium to insure that the fermentation proceeds at a reasonable rate (A8). Some of the limiting factors in the lactic fermentation were report- ed in 1928 by Rogers and Whittier (55). Although.Streptococcu8 lactis was used as the test organism, many of the observations apply to lactobacilli as well (3h, 35, 36). Control of the hydrogen ion concentration (pH) permits greater bacterial populations than when the accumulating acid is not neutralized. Even greater populations are attained when the culture with controlled pH is agitated with a mechanical stirrer, with air or with nitrogen, the effectiveness increasing in the order given. However, the concentration of undissociated acid is the principal factor in the limitation of growth and metabolism. A substance which is diffused through a semipermeable membrane, other than undissociated acid, also limits the growth of Streptococcus lactis (55)- More recently it was demonstrated that the yield as well as the rate of lactic acid production are functions of the pH of the fermentation. Up to an optimum value, the yield varies directly with pH; above the op- timum, an inverse relationship exists (12, 20, 31). A continuous process for the production of lactic acid from whey was developed by Whittier and Rogers (72). The process is unique in 'that sterilization of equipment and substrate are not required to prevent contamination by undesirable organisms. The fermentation temperature (US C) and pH (5.0 to 5.8) as well as the use of lactobacilli are suffi- cient to inhibit the growth of undesirable organisms. The continuous process was practiced to a limited extent on a commercial scale, but problems with residual sugar in the final product hindered the recovery of lactic acid (9). A.3 Cbntinuous Fermentation In general, continuous industrial processes are much more efficient than the corresponding batch processes (30, 69). This fact is adequately illustrated when one considers the chemical industry, in which most of the large-scale processes are operated continuously. Various authors report increases in productivity of five- to tenfold by converting batch fermen- tations to continuous fermentations (11, A0). This potential for very significant increases in productivity is reason enough to explore contin- uous fermentation when large-scale operations are anticipated. The primary factor which contributes to the increased productivity of continuous fermentations is the marked reduction in processing time with equipment of the same holding capacity (A2). A continuous fermenta- tion also is more adaptable to instrumental control, is better integrated into the preceding and subsequent processing operations, and generally yields a more uniform product. Many advances are being made in the de- sign of continuous fermentors. Some of the more novel approaches include plug-flow fermentors, tower fermentors, cyclone column fermentors (61) and dialysis fermentors (57). There are two potential problems which are unique to continuous fer- mentation: contamination and culture degeneration (AA). Contamination results from entry into the system of a fast-growing organism which even- tually displaces the original population. Culture degeneration results from genotypic mutations and selection or from phenotypic adaptations of the original culture. While these problems of continuous fermentation have prevented its widespread application, they have not proven insur- mountable. Many of the simpler fermentations such as the production of food and fodder yeast (A2, A9), ethanol (25) and vinegar (A, 65) have benefited from the use of continuous processes. More recently, novel processes for the production of single-cell protein from normal paraffins (58) and agricultural wastes (5) have also utilized continuous fermenta- tions. The successful introduction of a continuous process depends on a deeper knowledge of the process than is required for a batch process. Batch processes are often Operated as an art based on empirical knowledge. Continuous processes, on the other hand, require a fundamental knowledge of the microbiology, biochemistry and total kinetics of the process (A0). To gain a more fundamental understanding of fermentation processes,- it has been beneficial to describe the processes in terms of mathematical models. Equations have been developed which describe the relationship of throughput, microbial propagation, product formation, substrate utiliza— tion and the like. Due to the complexity of biological systems it is seldom possible to obtain a strict mathematical model of all the factors involved. Many excellent reviews discuss the theoretical analysis of con- tinuous culture systems (11, 18, A3, A6, 5A). Given a valid mathematical model for a continuous fermentation, it is possible to simulate the fermentation on an analog or a digital computer. It is possible to complete a computer-simulated fermentation in minutes whereas the actual fermentation may take hours or even days. Herein lies the value of the simulation: many process variables can be evaluated on the computer in a fraction of the time it would take to actually conduct the experiments. Furthermore, the cost of computer time is generally much less than the cost of the materials, equipment and labor used in the actual experiment. The extent to which an actual fermentation is pre- dicted by a computer simulation will depend on how closely the mathemati- cal model reflects reality (57). 5. MATERIALS AND METHODS 5.1 Bacterial Culture Lactobacillus bulgaricus strain 2217 (Chris Hanson's Laboratory, Milwaukee, Wis.) was used throughout this study. This organism was selected by Reddy (23) on the basis of its high rate of acid production in the pH range of 5.0 to 6.0. The culture was maintained in a sterile medium of 10% skim milk powder and 90% tap water contained in 25-ml screw-top test tubes. At least every two weeks the culture was trans- ferred to fresh medium. Inoculated tubes were incubated 18 to 2A hr at AA C. Coagulation of the milk served as a positive test for bacterial growth. After incubation the cultures were stored at A C. 5.2 Substrate Both fresh and reconstituted cottage cheese whey were used as fer- mentation substrates. Both types of whey were obtained from Michigan Milk Producers Association, Ovid, Michigan. It was determined experi- mentally that there was no difference in the two types of whey when either was used as the fermentation substrate. For the continuous fermentations, some difficulty was experienced initially in getting the powdered whey in- to solution. However, the following procedure was helpful in minimizing the problem. Three pounds of whey powder were put into a 5—gal polyethylene carboy along with 35 lb of water. The water and powder were partially mixed by putting the carboy on its side and rocking it for about a minute. This procedure wetted most of the powder, but many small lumps were still present. The carboy was cooled (A C) for 2A hr during which time the 10 11 lumps dissolved. Before use, the contents of the carboy were again mixed by rocking the carboy. 5.3 Fermentation Equipment All fermentations were carried out in a lA-liter bench-top fermentor with automatic temperature control (Model MAlAOFl, Fermentation Design, Allentown, Penn.). The pH was controlled by an automatic pH control module (Model pH-22, New Brunswick Scientific Co., Inc., New Brunswick, N. J.). Continuous fermentations were conducted by pumping reconstituted whey from an unsterilized feed reservoir to the fermentor and allowing the product to overflow into a product reservoir. The feed reservoir was a 5 gal, polyethylene carboy. This reservoir along with its contents were chilled (A C) for at least one day before using. While in use, the reservoir was insulated with a heavy blanket, which was sufficient to prevent the temperature of the whey from exceeding 16 C in 2A hr. The feed rate to the fermentor was controlled by a finger-type, peristaltic pump (Model T8, Sigmamotor Company, Middleport, N. Y.). Polyurethane tubing, obtainable from the pump manufacturer, was required for the section of line that passed through the pump; natural rubber tubing split in less than A8 hr, and Tygon tubing took a permanent set" that resulted in a variable flow rate. A simple overflow device was developed to maintain a constant level in the fermentor (Fig. l). A piece of l/A—in stainless steel tubing was coupled to one of the fermentor top fittings by means of a piece of l/A- in O.D. natural rubber tubing. The stainless steel tubing was cut to a length that determined the liquid level in the fermentor. A small amount of C02 was continuously purged into the fermentor and escaped through the l2 co * 3 2K . ) vwr ___-CI)- L___l FEED PRODUCT RESERVOIR PUMP FERMENTOR RESERVOIR h». . FIG. 1. Schematic diagram of'level-control device. Symbols: 1, stainless steel tubing; 2, rubber coupling; 3, balloon; 4, agitator. piece of stainless steel tubing, the only open port. A toy balloon taped to one of the fermentor top fittings served as a pressure release device in the event the outlet became plugged. As the liquid rose and covered the end of the stainless steel tubing, the 002 purge developed a slight pressure within the fermentor. The pressure was sufficient to force the liquid up and out of the fermentor. As the liquid was forced out, the liquid level dropped until 002 could again escape through the stainless tubing. The liquid level fluctuations were so minor as to be undetect- able to the casual observer. The liquid volume in the fermentor was 9.6 liters unless otherwise noted. Batch fermentations were conducted by inoculating 10 liters of unsterilized whey (pH 5.5, AA C) with 700 m1 of the pure bacterial culture. Reconstituted whey was prepared for the batch fermentations by adding suf- ficient tap water to 800 g of powdered whey to bring the liquid level in 13 the fermentor to 10 liters. The pH was maintained at 5.5 i 0.1 by the automatic addition of anhydrous ammonia. 5.A Procedure fer Continuous Fermentation To start, the continuous fermentor was charged with 10 liters of un- supplemented, reconstituted whey, the temperature and pH were adjusted to the desired levels, and 700 m1 of inoculum were added. The fermentation was allowed to proceed batch—wise until most of the lactose was fermented (approximately 2A hr). At that time the continuous feed was started. Samples were taken approximately every 12 hr, and the fermentation was allowed to proceed at least A8 hr before changing to a new set of operat- ing conditions. Other conditions were evaluated by changing the appro— priate parameters and allowing the fermentation to re-establish a new steady state. Normally, the feed rate was adjusted to give the reported retention times. However, to achieve the retention time of 7.6 hr in one stage and 15.2 hr in two stages, the liquid volume in the fermentor was reduced to A.9 liters; and the feed rate was adjusted accordingly. The continuous fermentor was in service for A2 days, when it was terminated voluntarily. The system was interrupted only for one weekend, at which time the fermentor and its contents were stored at A C. The fermentor was reinoculated at the time that it was put back in service, but this precautionary measure may not have been necessary. 5.5 Fermentor Measurements The pH of a sample from the fermentor was checked on a separate pH meter which was calibrated against buffers of known pH. The pH of the 1A sample was used to calibrate the pH control unit at least twice daily. The maximum error for the pH control system was estimated to be 1 0.1 pH unit. A thermometer inserted into the thermal well in the top of the fer- mentor was used to set the temperature control point. The temperature controller maintained the temperature within 1 0.5 C of the set point. The overflow rate from the fermentor was determined twice daily by measuring the collected product with a 2-1iter graduated cylinder. The flow rate varied less than 3% during the 12-hr collection period and was used to calculate average retention times. Samples were withdrawn through the top of the fermentor by means of a 25-ml pipette. The samples were quickly placed in the freezing compartment of a domestic refrigerator to stop further fermentation and held there until the time of analysis. 5.6 Lactose Determination Lactose determinations were made by a modification of the picric acid method of Perry and Doan (51). One ml of sample was pipetted into a 250-ml Erlenmeyer flask and diluted with 99 ml of saturated picric acid. If the sample was likely to contain less than 2% lactose, only A9 m1 of saturated picric acid were used; and the final result was divided by 2. The flask contents were shaken and filtered (#588 filter paper, Schleicher & Schuell, Keene, New Hamp.). Two ml of the filtrate were transferred to a 20 X 150 mm culture tube (which had been previously marked at the 20 m1 level) containing 1.0 ml of Na2C03 solution (25 g per 100 ml). The tube was stoppered lightly, shaken and placed in a boiling water bath for 20 1 0.5 min. The contents of the tube were then cooled to approximately 20 C in a water bath, diluted to 20 ml with distilled 15 water and mixed by inverting. A portion was transferred to a colorimeter tube, and a reading was obtained at 520 nm within 20 min of removal from the boiling water bath. A blank consisting of 2.0 ml of saturated picric acid and 1.0 ml of Na2C03 solution was heated, cooled and diluted along with the unknowns for adjustment of the zero point of the colorimeter (Model 20, Bausch and Lomb, Rochester, N. Y.). A standard solution was made by diluting 5.0 g of dry lactose to 100 ml in a volumetric flask. Aliquots of 0.2, 0.5 and 1.0 m1 of this solution were analyzed and the results were used to con- struct a standard curve. In no fermentation samples was an apparent lactose concentration of less than 0.2% obtained. It was suspected, therefore, that a nonspecific background color was present in the fermentation samples and was not pres- ent in the pure lactose samples used for constructing the standard curve. Four samples which had been reported as 0.2% lactose were submitted to the Department of Food Science and Human Nutrition at Michigan State Univer- sity for lactose analysis by the A.O.A.C. method (26). Two of the sam- ples contained no trace of lactose and the other two contained only a trace (less than 0.1%). These analyses confirmed that a nonspecific background color was in fact present in the fermentation samples. Conse- quently, the lactose results were corrected for this background color by subtracting 0.2% lactose from all samples which were analyzed with A9 m1 picric acid and subtracting 0.1% lactose from all samples analyzed with 99 m1 picric acid. Residual lactose was the preferred (even though indirect) indicator of rate and extent of fermentation because there was less variance in the lactose determinations than in the lactic acid determinations and because 16 the endpoint of the fermentation was indicated more precisely by lactose concentration than by lactic acid concentration. 5.7 Ammonium Ian Determination The ammonium ion concentration in the medium was used as an index of lactic acid production and was determined by a modification of a colori- metric method reported by Johnson (29). One ml of a sample containing A to A0 pg per ml of ammonium ion was pipetted to a spectrOphotometer cuvette. To the cuvette were added 2.0 ml of Nessler's reagent and 3.0 m1 of 2N NaOH. The Nessler's reagent contained (per liter) A.00 g of K1, A.00 g of HgI and 1.75 g of gum ghatti. The contents of the cuvette 2 were mixed by inverting and allowed to develop color at room temperature for 15 min. The absorbance was read at A90 nm. Blanks contained 1.0 ml of distilled water in place of the 1.0 ml sample. The standard for the ammonium ion determination was an (NHA)2SOA solution containing 100 ug ammonium nitrogen per m1. To prepare a stan- dard curve, 0.1, 0.3, 0.5, 0.7 and 1.0 ml of the standard and sufficient water to bring the volume to 1.0 ml were added to cuvettes in place of the 1.0 ml sample. 5.8 Lactic Acid Determination Lactic acid was analyzed by a simplified gas chromatographic proce- dure specifically and recently developed for bacterial metabolic prod- ucts (8). A 1.0 m1 sample of the culture was drained through 1 m1 of cation-exchange resin (Dowex 50W-X8, 50—100 mesh, H—form, washed in water; BioRad Laboratories, Richmond, Calif.) on glass wool in a Pasteur pipette. After the sample drained through the resin, the resin was l7 washed twice with 0.5 ml distilled water. All the fluid from the pipette was collected, and an aliquot was directly analyzed in a gas chromato- graph (Model 810, hydrogen flame detector, F&M Scientific Corp., Avondale, Pa.). A 1.8 m by 2 mm I.D. coiled glass column (Anspec Corp., Ann Arbor, Michigan) was packed with a porous polymer (Chromosorb 101, 80/100 mesh, Johns-Manville, Denver, Colo.). The column was conditioned overnight at 250 C and then run isothermally at 220 C. The inlet temperature was 250 C, and the detector temperature was 230 C. A 2.0 ul sample was in— jected. The carrier gas was 10 ml per min of nitrogen, and the hydrogen and air pressures were 7.5 and 11.0 psig, respectively. Solutions of 1.1, 2.2, A.3 and 6.0% lactic acid were analyzed, and the results were used to construct a standard curve. 6. THEORY A model specifically describing growth and product formation in the microbial production of lactic acid was developed by Luedeking (37), and this model is reviewed and modified below. This same model was published by Luedeking and Piret in a primary scientific publication (38, 39). Some of the data referred to as "Luedeking's data" in this thesis also may be found in the primary publication. A simple continuous fermentor is represented by the generalized model known as a continuous stirred tank fermentor (Fig. 2). It is FEED IN N! Av v —)- PRODUCT OUT Fig. 2. Continuous stirred tank fermentor assumed that such a fermentor is sufficiently mixed so that the composi- tion of the effluent is the same as the contents of the fermentor. A material balance of the bacterial mass of the system can be stated as: Rate of + Rate of = Rate of + Rate of Feed Production Withdrawal Accumulation l8 l9 Algebraically, the material balance is stated as: FXO + v(dx/de)G = FX + V(dX/d6) (1) where: F = rate of continuous feed, volume per unit time V = operating volume of liquid in the fermentor X = concentration of bacterial mass in the fermentor X0 = concentration of bacterial mass in the feed stream 9 = time (dX/de) = rate of change in concentration of bacterial mass due to growth, concentration per unit time. In Equation 1, (dX/de) is used to distinguish the rate of change in con- G centration of bacterial mass that is due to growth from the rate of change that is due to all other factors. (dX/de) represents the net rate of change from all factors. The specific growth rate k is defined as the rate of change in bac- terial mass per unit of bacterial mass present in the system: k = (dX/de) /x (2) G Dividing Equation 1 by V and substituting r for F/V results in the fol- lowing equation: rxo + (dx/de)G = rX + dX/de (3) where: r = throughput rate, volume of feed per unit time per unit operating volume. The reciprocal of r is the retention time of the average cell within the fermentor. When the feed is sterile (i.e., when X0 = 0) and the specific growth rate k is substituted into Equation 3, it becomes: dx/de = (k - r)X (A) 20 Under steady state conditions there is no change in the concentra- tion of bacterial mass as a function of time (i.e., dX/dO = 0). Therefore, the throughput rate is equal to the specific growth rate constant (i.e., I‘ k). This relationship is the basis for design of the classical Chemo- stat (or Bactogen) described by Monod (61). In the Chemostat, growth is limited by one of the nutrients or metabolic products. Consequently, the bacterial population adjusts itself until the specific growth rate is equal to the throughput rate. It is often desirable to relate the rate of product formation to some condition in the fermentor. Generally, the concentration of bacte- rial mass is used for this purpose. It has been empirically demonstrated that in lactic acid fermentations the rate of change of product concen- tration is a function of both the concentration of bacterial mass and also the rate of growth of the bacterial mass (37): (dP/de)G = a(dx/de)G + BX (5) where: P = concentration of product (dP/de)G = rate of change in concentration of product due to growth, concentration per unit time a, B = constants of proportionality fixed by the organism, substrate, pH and temperature. Since the specific growth rate is given by (dX/dO)G/X, Equation 5 becomes: (dP/de)G = (dk + 8)X (6) Therefore, a material balance for the product can be written as follows: dP/de = (ak + B)X — rP (7) 21 For homofermentative lactic acid fermentations it can be assumed with reasonable accuracy that the rate of product formation is proportional to the rate of substrate utilization (12, 20, 37), because only a small, relatively constant fraction of the substrate is incorporated into bacte- rial mass (20). Therefore: (dP/de)G = —Y(dS/dG)G (8) where: S = concentration of substrate (dS/de)G = rate of substrate utilization by the bacterial culture Y = yield constant expressed as the ratio of product formed to substrate consumed. A material balance for the system again yields an expression for the rate of change in the substrate: dS/de = (So — S)r — (ak + B)X/Y (9) where: S0 = concentration of substrate in the feed. In summary, the material balance equations for bacterial mass, prod- uct and substrate in continuous lactic acid fermentations are: dX/dO = (k - r)X (A) dP/de = (at + R)x - rP (7) dS/de = (SO - S)r - (ck + B)X/Y (9) It should be noted that no assumption has been made for a constant value for k as was done previously (37). As a result, the above equa- tions should hold both for the logarithmic growth phase where k is a con- stant and for the phase in which an accumulation of metabolic products causes a decline in the value of k. 22 To express the specific growth rate as a function of the product concentration, assume that the following linear relationship exists: ki = kmax(l - P/Pmax) (10) where: k1 = specific growth rate for a given product concentration if sufficient substrate were present. k = maximum specific growth rate attained in max . the fermentation. Pmax = product concentration at which ki first equals zero. 0 Equation 10 is a simple relationship between ki and kmax which adequately describes the data of Luedeking in the pH range of A.8 to 5.6 (37). By inspection, one can see that Equations A, 7 and 9 are not valid when the substrate has been exhausted. Equation A would predict an in- crease in bacterial mass even when there was no substrate present. Simi- lar problems can be observed in Equations 8 and 10. In enzyme kinetics, the Michaelis-Menten equation is used to relate the enzyme reaction rate to the concentration of substrate: V S v _ _E@§L_. K + s (11) s where: v = enzyme reaction rate V = maximum enzyme reaction rate if substrate max . . . were not limiting. Ks = saturation constant Monod found that bacterial growth was analogous to an enzyme reaction (11). Therefore the reaction rate can be described as a function of sub- strate and metabolic products by incorporating Equations 10 and 11 into 23 dX S 33 = Ei (Ks + S) " TX (12) In Equation 12, k1 is analogous to Vmax in Equation 11. In Equation Equation A to give: 7, both Ok and B are reaction velocities which must be dependent on substrate concentration. Again, use of Equation 10, analogy with Equation 11 and application of both to Equation 7 give: gg _ S de - (a1:i + s)(———-———KS + S)x — rP (13) Similarly, Equation 9 becomes: dS _ S X d—e- — (So - S)I‘ - (Gki + 8) (————KS + S)? (1)4) Therefore, the revised material balance equations which take into account a variable specific growth rate constant and substrate exhaus— tion are: dX -—' = k. S (16 1 m — r X (12) _ S 9.11 — s d6 - (dki + B) (m) X - I‘P (I) dS _ S 25 (1A) ”(i—e- - (SO _ S)I‘ " (Gk + 8) (KS + S) Y By inspection one can now see that Equations 12, 13 and 1A could be valid even when the substrate is exhausted. As S approaches 0, the quan- tity S/(KS + S) must also approach 0, indicating that no more change can take place by bacterial metabolism. 7. RESULTS 7.1 Cbmputer Simulation of Batch Fermentation The material balance equations (Equations l2, l3 and 1A) served as the basis for the computer simulation, which was programmed on a digital computer by Dr. George Coulman of the Department of Chemical Engineering, Michigan State University. The purpose of the simulation was to guide the experimental part of the research program. Therefore, it was neces- sary to execute the program before the experimental results were avail- able. However, some first approximations were required for the constants in the material-balance equations. These initial values were taken from the data of Luedeking (37) with full realization that different media, cultures and neutralizing agents were used in the two systems. However, these differences were inconsequential since the purpose of this phase of the programming was to indicate general trends and not to yield absolute values. To verify that Equations 12, 13 and 1A and the constants were in fact valid, they were first used to simulate Luedeking's data for batch fermentation at pH 5.6. The program was started at Hour 3 to avoid the problems associated with simulating the lag phase. This was no compro- mise in the original objectives since a lag phase does not occur in con- tinuous culture. The values of the various constants and the initial conditions are given in Table l. The specific growth rate constant ki given by Luedeking varied throughout the fermentation (Fig. 3). In the pH range of A.8 to 5.6, ki was approximately proportional to the product concentration. Therefore, Equation 10 was used to approximate the curve for pH 5.6. For P > 38, 2A TABLE 1. Values used in the simulation of batch fermentation, from Luedeking (37) Symbol Value Reference a 2.2 (37), p. I—24 8 0.49 " " P (initial) 1.1 " Table I—l, Hour 3 S (initial) 50. " " " X (initial) 0.283 " " " for P 538: 0.48 Fig. 3 in this thesis Pmax 50 H II N H H max ' for P >38: k 1. l H H H II N Pmax 43 n n n n n max ' r 0. (Batch fermentation) 26 0.5- : 0|” -4 PE :2" t”: 0.3 .4 SE S 55 0.2 _. U E: 23 '5': C0 Oll '- 0.0 1 1 1 ‘ x 0 10 20 30 AD 50 LACTATE CONCENTRATION (MG/ML) FIG. 3. Effect of lactate concentration of the specific growth rate, zaith data from Luedeking (37). A: pH 4.8. symbols: as, pH 5.6; 0, pH 5.4; a, pH 5.2; 27 the slope of the curve became more negative; therefore, the values for kmax and Pmax were reset to reflect this change (Table l). The mathematical model (Equations 12, 13 and 1A) was shown to simu- late closely Luedeking's batch fermentation data (Fig. A). Near the end of the fermentation (i.e., Hour 13 to 1A), the simulated bacterial densi- ty deviated from the experimental bacterial density because the simulated product concentration was slightly less than the experimental product concentration. Therefore in the simulation, a small amount of substrate remained after 13 hr to support bacterial growth. 7.2 Simulated and Experimental Centinuous Fermentation After Luedeking's batch fermentation data were found to verify the mathematical model, the same model was modified and used to simulate single- and multi—stage continuous fermentations. As indicated in Table l, the term r in Equations 12, 13 and 1A was set to 0 for the simulation of a batch fermentation. However, this term was retained when simulating a continuous fermentation. In simulating a multi-stage continuous fermen- tation, the product from one stage was used as the feed for the subse— quent stage. Equations 12 and 13 were then modified to take into account the fact that an inlet stream to a given stage could contain appreciable amounts of product and bacterial mass. Equation 1A previously contained a term for the substrate concentration in the inlet stream. Therefore, the three generalized material-balance equations for any stage of a multi- stage fermentor are: u d8 II P A 74 U) +0) U) V I Q >4 + '1 >4 0 : \J'l g: = (0.1.1 . B)(-K-——E——§>X + r(PO — P) (16) 28 50 AD 30 BACTERIAL DENSITY (U.O.D./NL) 10 16 TIME (HR) FIG. A. Camparison of curves from the computer simulated data with the experimental data of.Luedehing (37) fer bacterial density (X) and lactate concentration (0). LACTATE CONCENTRATION (MG/ML) dS _ S X (17) The concentrations X0, Po and So refer to the concentration of the bacterial mass, product and substrate, respectively, in the inlet stream to a given stage of a fermentor. Equations 15, 16 and 17 were used to simulate continuous fermentors with one, two and three stages. As in the batch fermentation, Luedeking's data at pH 5.6 was used to define ki at various product concentrations (Equation 10 and Table l). The feed to the multi—stage continuous fermentors contained 50 mg per ml glucose, no bacteria and no product. The results of the simulation (Fig. 5) indicated the general trends one could expect by varying the retention time and by using a multi-stage fermentation. The total retention time for all stages is given in Fig. 5. The retention time for each stage can be calculated by dividing the total retention time by the number of stages. The ordinate in Fig. 5 refers to the amount of glucose in the effluent from the last stage. As the reten- tion time in a simulated, single—stage fermentation was increased, the concentration of residual sugar decreased. A point (at approximately 10 hr) was reached at which only marginal improvements were achieved by fur- ther increasing the retention time. Further reductions in the concentra- tion of sugar were achieved by separating the fermentation into two and three stages. However, three stages gave only marginal improvement over two stages. The experimental results for the continuous lactic acid fermentation of reconstituted whey (Fig. 6) were similar to those for the simulated glucose fermentation (Fig. 5). The experimental fermentation at pH 5.5 reached a point (at approximately 15 hr) at which a further increase in 3O F l I I 50 . ._ LID _ I-Stage _) 2-Stages 30 ._ ._ M 3-Stages RESIDUAL GLUCOSE (MG/ML) 20 ._ -a 10 - . - 0 (— ‘ XthX— —-7 I l l J 0 5 10 15 20 25 TOTAL RETENTION TIME (HR) FIG. 5. Effect of total retention time in single- and multi-stage fermentations as predicted by simulation. Symbols: 0, one stage; X, two stages; A, three stages. 31 [, I l l I T7 l 5 .. ?'-~‘ ’ \ I I u - I _. I I 13 ‘ \ .’ ‘3, 3 _. ‘ .. '68 I E I 5 t gg __ <3 c: 2 '3 83 32' \ 1 - \ .. \b To 0 o g::‘ <—~o -30 ‘\\~ I I l I l l 0 5 10 15 20 25 30 35 TOTAL RETENTION TIME (HR) FIG. 6. Effect of the total retention time and staging on the level of residual lactose in continuous fermentaion of'reconstituted whey. The results shown in Fig. 5 were used to interpolate the curve for the single-stage (0) between 0 and 7 hr and fer the double-stage (X) between 12 and 31 hr. 32 the retention time caused essentially no further reduction in the residual lactose concentration. However, the use of two stages significantly re- duced the level of residual lactose. In fact, it was possible to achieve more than 98% lactose conversion by using two stages with a retention time of 15.5 hr in each stage (i.e., a total retention time of 31 hr). Sufficient information is given in Fig. 6 to extend the mathematical model to the whey fermentation. One can obtain ki as a function of P (Fig. 7) by recalling that ki is equal to the reciprocal of the retention time under steady state conditions as long as S is much greater than KS (Equation 12). Therefore for the whey system under steady—state condi— tions (i.e., dX/dO = 0), there were no unknowns in Equation 12. Since and B are not independent, Equation 13 contained only two unknowns: (OLki + B) and X. Hence, Equation 13 (and similarly Equation 1A) was solved for X by assigning an arbitrary value to d and adjusting B by successive approximations to fit the model to the experimental data. The previous values of 2.2 and 0.A9, respectively, were used as first approximations for a and B. By successive approximations, values of a = 2.2 and B = 0.2 were found to yield a good fit of the model to the experimental data (Fig. 8). Therefore, it was possible to extend the :model to include the continuous fermentation of whey. The model then en— abled one to predict the performance of a single- or multi—stage fermen- 'tor Operating over a wide range of retention times. 7.3 Adaptation of the Continuous Culture At the end of the A2 day continuous fermentation, the bacterial cul— ‘bure reduced the lactose concentration to a lower level than at the start. 'This phenomenon is illustrated by the data in Table 2. In the early part <3f the fermentation, with a pH of 5.5 and a retention time of 22 hr, the 33 0.5 O .1: From Luedeking's data at pH 5.6 ‘Ar’:E;/’ (see Fig. 3) From whey experimental data at pH 5.5 (see Fig. 6) c: L»: I c: R: .1 SPECIFIC GROWTH RATE. K, (HR'l) P H I PRODUCT CONCENTRATION. P (O/o) FIG. 7. Graphs of the linear approximation used to simulate the effect of product concentration on the specific growth rate with calculated data points indicated (X). 3A I II I I l l 5 _ S h T-Stage ii p _. Jkrqu””’I -2-StageS ,‘ ‘ur”2—”””” \0 C31 3 _ 3-Stages _ 5,4 W (3 t3 :5 a fig 2 L. ,_ 33 32' l - .. . To 1‘ #453 _:x>- 0 *- \\—~2—- xx... ..., 0 l I I I I I 0 5 10 15 20 25 3O 35 TOTAL RETENTION TIME (HR) FIG. 8. Effect of the total retention time and staging on the level of'residual lactose in whey as predicted by computer simulation (curves). Experimental data points fer single-stage (0) and dauble-stage (X) fer- rnentations are included fer comparative purposes. 35 TABLE 2. Adaptation of bacterial culture as indicated by a change in residual lactose concentration during prolonged continuous fermentation at two levels of pH Retention Residual pH Day Time (hr) Lactose (%) 2 22.0 1.9 3 21.6 1.7 11 21.4 1.6 5.5 12 21.6 1.6 39 22.2 0.5 40 22.6 0.6 6 22.0 1.0 6 22.2 1.0 17 15.7 0.9 5.8 17 14.7 0.8 18 14.5 0.7 19 11.7 0.6 19 12.0 0.7 36 culture reduced the lactose level in the whey to about 1.7%. A month later, at the same operating conditions, the culture reduced the lactose level to about 0.6%. The steady improvement in performance also occurred at pH 5.8. Al- though the retention time was reduced over a 13 day period, the percent- age of lactose in the product was also reduced during that time period, in contrast to the results presented in Fig. 6, A qualitative observation also indicated a change in the bacterial culture. The L. bulgaricus #2217 culture used to inoculate the continu- ous fermentor on Day 1 had a pronounced, yoghurt-like odor. However, an inoculum prepared from a freeze—dried sample of the continuous fermen— tor culture taken on Day 35 had very little odor of any kind. The data in Table 2 were used to identify the period from Day 1A to Day 19 as a period of significant change in the qualitative ability of the culture to ferment lactose to lactic acid. Therefore, no data from this time period were used to illustrate the effect of pH or retention time on the conversion of lactose. 7.A Effect of'pH in the Range of 5.5 to 6.0 on Continuous Fermentation Increasing the pH of the continuous fermentation from 5.5 to 6.0 resulted in a significant reduction in the amount of residual lactose in the final product. The data in the upper portion of Table 3 show the ef- fect of pH before the adaptation of the bacterial culture. Since there was no significant difference in the retention times (in the upper portion of the table) the difference in levels of lactose was due to the pH effect. A 37% reduction in the amount of residual lactose (from 1.70% to 1.07%) occurred when the pH was raised from 5.5 to 5.8. Further increasing the pH to 6.0 only reduced the residual lactose concentration an additional 7% (from 1.07% to 0.95%). 37 TABLE 3. Effect of'pH in the range of 5.5 to 6.0 as indicated by resid- ual lactose concentration Residual lactose (%) Retention time (hr) Period pH Replication Average Replication Average 1.9 22.0 1.7 21.6 5.5 1.6 1.70 2l.A 1.6 21.6 Before 1.0 22.0 Ada tation 5.8 1.0 1.07 22.2 21.3 ‘ p ' 1.2 20.0 0.9 20.6 6.0 1.0 0.95 20.3 0.7 15.7 0.7 15.A 5.5 0.8 0.70 15.2 0.7 15.2 0.6 15.2 15.2 0.3 1A.8 After 6'0 0.3 0'30 15.1 Adaptation 1.1 11.5 5'5 1.1 1'10 11.5 11.3 0.8 11.0 6.0 0.8 0.80 11.3 38 In the lower portion of Table 3 is shown the effect of pH at two dif- ferent retention times for the period after the adaptation of the bacte- rial culture. Increasing the pH from 5.5 to 6.0 resulted in 57% (from 0.70% to 0.30%) and 27% (from 1.10% to 0.80%) reductions of residual lac- tose at retention times of 15.2 and 11.3 hr, respectively. 7.5 Quality of.Product from Continuous Fermentor The odor of each sample was checked to see if there was any indica- tion of putrification, and at no time was this noted. The color of the product from the continuous fermentor was the same as that of the whey fed into the fermentor. This was in contrast to the batch fermentations, in which the color of the whey changed to a light brown as the fermentation neared completion. The fermented product was analyzed by gas chromatography to check for the presence of products of metabolism other than lactic acid. Chromato— grams of the feed and product were compared to a chromatogram of a stan— dard solution which contained A.A% lactic acid plus 0.1% acetic acid and 0.2% ethanol (Fig. 9). The relative amount of acetic acid and ethanol in a fermentation sample was determined by comparing the height of the re- spective peaks to the height of the background peak, which appeared with all samples including pure water. Peaks which appeared at the same reten- tion time as acetic acid and ethanol were assumed to be acetic acid and ethanol peaks although a positive identification of the experimental peaks was not made. The assumption is believed to be a good one, since hetero- fermentative lactic acid bacteria producing CO2, acetate and ethanol could also propagate at the restrictive conditions within the continuous fermentor. In Fig. 10, the upper chromatogram is from the product after A2 days of continuous Operation, and the lower one is from a feed sample. These 39 N _) --... __‘_ FIG. 9. Gas chromatogram of'4.4% lactic acid standard with addi- tion of 0.2% ethanol and 0.1% acetic acid (attenuation = 16). Symbols: 1, background; 2, ethanol; 3, acetic acid; 4, lactic acid. A0 S A 3 N '5 FIG. 10. Gas chromatograms of'product from (top) and feed to (bottom) continuous fermentor (attenuation = 8). symbols: 1, background; 2, ethanol; 3, acetic acidk 4, lactic acidk 5, unknown material. Al chromatograms are typical of those obtained from samples taken throughout the course of the experiment. In both chromatograms, traces of ethanol and acetic acid are indicated as well as traces of two unknown compounds which elute just before and just after lactic acid. The concentrations of acetate and ethanol were greater in the product than those in the feed although they were less than the 0.1% and 0.2%, respectively, shown in Fig. 9. The concentration of the unknown compound eluting before lactic acid was substantially less in the product than in the feed, and the con- centration of the unknown compound eluting after lactic acid was about the same in the product as in the feed. Compounds that elute between acetate and lactate were not produced in the fermentation, including butyric acid which had been found to elute about one-third of the way between acetate and lactate. It was concluded, therefore, that lactic acid was in fact the predominant product of the long—term continuous fermentation and that a significant number of butyric acid-producing organisms were not propagated in the continuous fermentor. All of the above experimental results were obtained with a continuous fermentation lasting A2 days. A preliminary continuous fermentation last- ing 15 days also produced a high—quality product and essentially confirmed the above effects of pH and retention time on the concentration of resid— ual lactose. However, mechanical problems with the feed pump and the ab- sence of a discernible adaptation of the culture prevented a quantitative comparison of the two continuous fermentations. 7.6 Effect of'Yeast Extract and Cornsteep Liquor on Batch Fermentation Two series of three batch fermentations each were made to determine whether growth factors would improve the rate of fermentation in fresh whey. Two crude sources of growth factors were used: yeast extract at A2 levels of 0.1, 0.2 and 0.A%, and cornsteep liquor at levels of 0.25, 0.50 and 1.00%. The cornsteep liquor was autoclaved at 121 C for 30 min prior to use. The whey for all three fermentations in a series was from the same lot. Both the ammonium ion concentration and lactose concentration were used to determine the rate of fermentation. Whey supplemented with yeast extract was fermented more rapidly and to a greater extent than unsupplemented whey. In Fig. 11 is shown the effect of this growth supplement on the accumulation of ammonia as a func- tion of time. A typical unsupplemented fermentation is presented for com- parative purposes. The unsupplemented fermentation was conducted on a separate lot of whey which contained less lactic acid at the start of the fermentation, hence the initial level of ammonia was slightly lower than in the other three fermentations. The 25 hr fermentation time was typical of the many fermentations which were conducted before it was discovered that growth factors significantly improved the rate of fermentation. Lactic acid and residual lactose also were analyzed in samples from the series supplemented with yeast extract. The results are shown in Fig. 12 and complement those in Fig. 11. Yeast extract levels of 0.2 and 0.A% gave better results (i.e., they reduced the lactose concentra— tion to a lower level) than 0.1% yeast extract. The higher concentrations of yeast extract reduced the fermentation time to approximately 12 hr as compared to the 2A to 30 hr encountered when no supplements were used. The residual lactose analyses for the unsupplemented fermentation were not available for comparison. However, the data for the 0.1% level of yeast extract in Fig. 11 and 12 and the level of NHA+ in the unsupplemen— ted control in Fig. 11 indicated that considerable lactose (0.5 to 1.0% remained in the product when no supplements were used. A3 x’: 0.2% ' - C) AMMONIUM ION CONCENTRATION (MG/ML) 0, I 1 l I l O 5 10 15 20 25 FERMENTATION TIME (HR) FIG. 11. Effect of three concentrations of yeast extract on the rate of batch fermentation of fresh whey, measured indirectly by ammoniwn ion concentration. Symbols: :3 , 0.1%; O, 0.2%; X, 0.4%; O, typical unsupple- mented fermentation. AA 7 I T I I 7 0 6h- -5 O 5 -5 A \I\ 5 OS ' . 0.2% 5 ~a :3 :5. 4— - —A E E X- 8 275’ E U U a 3" ‘3; 5 . . 0.1% g 5 [/0195 2— x ' 42 .- 0.A% - x - l— a --1 \ I 0 " ‘xo-- - O J I I l O 5 10 15 20 25 TIME (HR) FIG. 12. Effect of three concentrations of’yeast extract on the rate of’batch fermentation of‘fresh whey as measured by the appearance of'lactic acid (closed symbol) and the disappearance of'lactose (open symbols). AS A material balance on one of the batch fermentations indicated that no significant portion of the substrate was lost to products other than lactic acid (Table A). Table A is a list of the concentrations of lac- tose and lactic acid at various times during the fermentation. The last column in the table is the sum of the two concentrations. The totals found in the last column are within 1 10% of the mean. The variance is random, and the totals do not steadily decline as would happen if a detectable amount of substrate was lost to CO2. A second observation indicated the lack of gaseous products: the batch fermentations were con- ducted in a closed fermentor with a toy balloon as a safety device for re- leasing any excessive gas produced by the fermentations, and at no time was the balloon extended due to gas production. In Fig. 13 is shown the effect of three levels of cornsteep liquor on the rate of fermentation. One percent cornsteep liquor gave only mar- ginal improvement over 0.25 and 0.50% cornsteep liquor. A comparison of Fig. 12 and 13 reveals that cornsteep liquor at a level of 0.25% per- formed nearly as well as 0.2% yeast extract when the two were used as sources of growth factors in fresh whey. 7.7 Ammonium Lactate and Calcium Lactate Inhibition of Batch Fermentation Equation 10 is based on the assumption that the specific growth rate k1 is diminished as a linear fUnction of the product concentration. The data of Luedeking, given in Fig. 3, support but do not validate this assumption. In fact, a plot similar to Fig. 3 of ki vs. substrate (rather than product) concentration would indicate that there also is a linear re- lationship between ki and substrate concentration. To resolve whether in fact product inhibition or substrate exhaustion had the predominate effect on ki’ a series of batch fermentations were conducted in which an excess A6 TABLE 4. Material balance data fer a batch fermentation of’fresh whey supplemented with 0.2% yeast extract Fermentation A B (A+B) time (hr) Lactose, % Lactic acid, % Suma, X 0.0 4.8 1.3 6.1 0.8 4.7 1.2 5.9 1.8 4.5 1.4 5.9 3.5 3.7 2.5 6.2 4.5 2.6 3.8 6.4 6.3 1.5 4.5 6.0 7.5 1.1 4.5 5.6 8.6 0.7 5.3 6.0 9.5 0.4 6.5 6.9 10.5 0.3 5.9 6.2 11.5 0.1 6.2 6.3 21.0 0.0 6.8 6.8 a Mean = 6.2; standard deviation = 0.4 h? 6 7 l l I l 5 E _ >8 A M a u — ‘c _ 9, E E .... g 3 - ... § 9 ‘64 E x S 2 .. \A .. x \ 0.25% and 0.50%. /‘7 o A 1.0% o l _. _ x \1\°~o 0 — \x~x>‘a* - J I I l J 0 S 10 15 20 25 TIME (HR) FIG. 13. Effect of three concentrations of'cornsteep liquor on the rate of’batch fermentation of'fresh whey. Symbols:‘4 , 0.25%; O, 0.50%; x, 1.0%. h8 of substrate was always present and varying amounts of product were pres- ent. If each batch fermented to the same extent, it would indicate that k1 is independent of product concentration. However, if each batch fer- mented to an extent which was inversely proportional to the amount of product present, it would indicate that k1 is dependent on the product concentration. In the experiment, the terminal pH was used as a measure of the extent of fermentation. That is, the terminal pH indicated the extent of fermentation before ki was reduced to 0. Both ammonium and calcium lactate were evaluated as potential in- hibitors by adding h ml of a bacterial culture and varying amounts of each lactate solution to ho ml of reconstituted, unsterilized whey. The whey had been previously neutralized to pH 7.0 with 30% aqueous NH3. The lactate solutions were prepared by neutralizing reagent—grade lactic acid with either ammonium hydroxide or a slurry of calcium hydrox— ide. The neutral calcium lactate solution contained hh% (w/v) lactate ion, and the neutral ammonium lactate solution contained 59% (w/v) lac- tate ion. The salt solutions were standardized at hh% (w/v) lactate ion by diluting the ammonium lactate with distilled water. The calcium lac— tate solution was a solid at room temperature and was liquified for trans- ferring by heating in a boiling water bath. The following formula was used to calculate the volume of lactate solution to be added to the whey samples: ho ml whey 1 x g lactate l 1.00 ml solution 100 ml whey A O.hfl g lactate Y (15) where: X desired percentage of lactate supplementation K: II volume of solution to be added, ml. to Two bacterial cultures were used. One was L. bulgaricus #2217 which had been stored in a stab culture for five months in a refrigera- tor, during which time the refrigerator malfunctioned for about two weeks. Apparently, the fermentative capability of the culture changed dur— ing storage as the culture did not ferment lactose as vigorously as the L. bulgaricus culture used for the other parts of the experiment. The changed condition of the culture did not affect this comparative study. The second culture was obtained from a freeze-dried sample of the insolubles recovered from the continuous fermentor on the 3hth day. Both cultures were propagated in sterile, 10% skim milk. The lactate supplemented whey cultures were incubated at h2 C. After 17 and hl hr of incubation, the pH of each sample was determined with a pH meter. Both ammonium lactate and calcium lactate inhibited the whey fermen— tation, although the former was significantly more inhibitory (Fig. 1h). The mixed culture was more acid tolerant than the #2217 culture. The mixed culture lowered the pH of the unsupplemented samples to about 3.9 whereas #2217 culture lowered the pH to 4.6. At all levels of supple— mentation the mixed culture was able to reduce the pH to values lower than those achieved by the other culture. 50 6.0 Reference line (see section 8.6) dI/I£T"">--"--(h~..3“‘w3 505 ’4‘:"&_ 5.0 _- figu://$kfiflfiyaaa—vvvt’ ” -. 7" / 45 PH l \ \t ‘43,, \ f ‘\ D ‘a 4.5 ‘0‘} I: d ' (’30 .— AF’ “A*° 4"’ I I ,aa’q" :30' «0" a" a” L100 -/,$‘" .- 3.5 J l l l l. 0 2 4 6 8 10 LACTATE ION SUPPLEMENTATION (0/0) FIG. 114. Effect of ammoniwn and calcium lactate on the terminal pH of batch fermentations with L. bulgaricus #2217 or a mixed culture of lactic acid bacteria. Symbols: 0, culture #2217 with ammonium lactate; X, culture #2217 with calcium lactate; A, mixed culture with ammonium lactate; a, mixed culture with calcium lactate. 8. DISCUSSION 8.1 Effiect of'Retention Time and Staging on conversion of’Lactose A comparison of the results from the batch fermentation model and the experimental data (Fig. 4) shows how well the model fits the data. The model, therefore, is a good simulation of the batch data. Care, however, must be exercised in applying the batch model to a continuous system (20,62). Due to product inhibition, many of the kinet- ic advantages of continuous fermentation are not fully realized in the lactic acid fermentation. Approximately the same time was required to process a given volume in the continuous fermentor as in the batch fer- mentor. However, the use of staging and increasing the pH would reduce the required retention time in the continuous fermentor. In spite of the fact that the retention times in a batch and continuous fermentor may be approximately the same, continuous fermentation would likely be the pref- erable mode of operation since it is more adaptable to instrumental con- trol, is better integrated into the preceeding and subsequent processing operations, and generally yields a more uniform product. Equation 12 predicts that the bacterial population will increase as long as the throughput rate r < kiS/(Ks + S). In a fermentation in which the specific growth rate ki is a constant, the population will increase until the substrate S is nearly exhausted. Therefore, the fermentation al- ways operates near the point of substrate exhaustion. If r > kiS/(Ks + S), the bacterial culture is washed out of the fermentor. In the lactic acid fermentation, however, ki is not a constant but rather varies as some function of product concentration (Fig. 7). As r 51 52 is decreased (i.e., retention time is increased), the product concentra- tion increases and causes ki to decrease. Therefore, further increases in retention time are counteracted by a reduction in the specific growth rate. Hence, the curve in Fig. 5 flattens, and very little improvement in lactose conversion is achieved by increasing the retention time beyond the optimum. The second stage of a two-stage fermentation does not depend on bac- terial growth alone for maintaining the population within the second fer— mentor. The second stage receives a supply of bacteria from the first stage. Therefore even if k1 is zero for the second stage, there will be bacteria in that stage as long as r < kiS/Ks + S) in the first stage. The bacteria that are fed to the second stage can use maintenance metabo- lism to dissimilate lactose even if k1 = 0. This fact, along with a low concentration of lactose in the feed to the second stage, account for its effectiveness. It should be pointed out that ki has been presented as a function of product concentration and not of substrate concentration. It appears that a fermentation with an initial substrate concentration of less than 5% would benefit relatively little by staging (i.e., the curves in Fig. 6 would be shifted downward). Conversely, a fermentation with an initial substrate concentration in excess of 5% would receive relatively more benefit from a multi-stage fermentor than demonstrated in Fig. 6. From a practical standpoint, this would indicate that cheddar cheese whey (h.9% lactose, 0.2% lactic acid) may do adequately in a singleestage fermentor, while cottage cheese whey (5.8% lactose, 0.7% lactic acid) may benefit from staging. Whey fortified with additional solids from con- densed whey may also benefit significantly by staging. The addition of molasses or other concentrated substrate to whey to increase the sugar 53 concentration may also benefit by a multi-stage fermentor. The purpose of adding the concentrated substrates to whey would be to reduce the amount of water that must be removed to get a pound of concentrated prod— uct. For example, the addition of enough molasses to whey to double the substrate concentration would cut in half the amount of water that must be removed to get a pound of product. Once a valid mathematical model of the lactic acid fermentation is available, it can be used to evaluate not only various substrate concen- trations (as outlined above) but also various fermentor designs. Plug- flow fermentors with a portion of the product used as a continuous inocu— lum (2h), multi-stage fermentors with stages of varying sizes, dialysis fermentors (57) and ultrafiltration fermentors (62) are a few systems which could be evaluated with the simulation. In each case experimental results must be used to verify the model and to determine if the culture shifts to a different metabolic pathway (62). In this study, the model aided in the interpretation of the limited data from the two-stage fermentor and predicted the performance of a three-stage fermentor (Fig. 8). The experimental data indicate that a lactose conversion of more than 98% can be achieved with a retention time of 15.5 hr in each of two stages. The model predicts that it may be pos— sible to achieve the same conversion with a retention time of ll hr in each of two stages or 5.5 hr in each of three stages. To fit the model to the experimental data it was necessary to find the relative values of the proportionality constants a and B. It should be noted that the relative values of a and B (i.e., the ratio of a to B) and not the absolute values were critical to fitting the model to the experimental data. To demonstrate that other values of a and 8 could Sh also force the model to fit the data, values of 2d and 28 were sub- stituted for a and B. The resulting product and substrate concentra- tions remained the same as when d and B were used in the program; how— ever, the apparent bacterial density was reduced to one-half the pre— vious value. The true values of a and 8 could be found if the true bacterial density were known at any one retention time. For the purposes of this study the relative values of a and 8 served just as well as the true values, since the bacterial density was of no particular interest. An- other reason for ignoring bacterial density was the difficulty in obtain— ing a value for it in a turbid medium such as whey. Plate counts are of little value due to the tendency of lactobacilli to form chains. DNA composition of the fermented medium may be an indirect method of deter- mining bacterial numbers since each cell contains a fixed amount of DNA (17). The relative values of a and B can be used to determine the rela- tive contributions of bacterial growth and maintenance metabolism to lac- tic acid production. For example, in a single-stage whey fermentation with a retention time of 5 hr, the residual lactose concentration is h.2% (Fig. 8); the product concentration is 1.6% (1.0% in feed + 0.6% produced in the fermentor); and ki equals 0.2 (Fig. 7). Since 0 k1 accounts for product formation due to bacterial growth and 8 accounts for product formation due to maintenance metabolism, the respective contributions are 0.hh and 0.20 when a = 2.2 and B = 0.2 Therefore, in this situa- tion 70% of the lactic acid is formed due to bacterial growth, and 30% is formed due to maintenance metabolism. In the same fermentor with a retention time of 30 hr (i.e., ki = 0.03), a ki = 0.07 and B = 0.2. 55 Therefore in this situation, only 25% of the lactic acid is formed due to growth, and 75% is formed due to maintenance metabolism. 8.2 Adaptation of'the continuous Culture The data in Tables 2 and 3 indicate that a definite change took place in the culture during the A2 days of continuous fermentation. This change has been loosely referred to as an "adaptation". It was never elu- cidated, however, whether in fact this was the selection of a mutant, the replacement of culture #2217 by another lactic acid bacterial species or a symbiotic association of #2217 with a mycoderm or other bacterial species. Any one or a combination of the above changes could have taken place in the culture. Whatever the actual change, it was for the overall benefit rather than detriment of the whey fermentation. After the adaptation, it was possible to increase the throughput rate by a factor of 3 and still achieve the same degree of conversion of lactose that was achieved at the lower rate before adaptation. For example, before the adaptation a reten- tion time of 22 hr and a pH of 5.5 resulted in a residual lactose level of 2.1%. After the adaptation a retention time of 7.6 hr at the same pH resulted in a residual lactose level of 2.3%. The adaptation period appears to have been a discrete time period. After that period the culture was stable, and the results were reproducible. It is possible that prolonged operation of the continuous fermentor would result in the selection of a third culture which was even more efficient than the culture which was present at the time the fermentation was ter- minated. The adaptation of the bacterial culture supports the argument that a continuous fermentor must be operated for an extended period of time 56 before much significance can be applied to the results. It has been arbi- trarily stated that a continuous operating time of 1000 hr should be achieved before a fermentation system is called "continuous" (61). This is proposed because so many changes can take place in a continuous fer- mentor, and only time will determine if those changes will in fact affect the system in question. The use of unsterilized medium and equipment make it very easy for contaminants to enter the fermentor. Any time a faster growing species enters, it will displace the predominate species. However, the restric- tive conditions in the fermentor (i.e., low pH, high temperature, anaero- bic conditions, lactose substrate and possible antibiotics produced by the lactobacilli) make it very unlikely that species other than lactic acid bacteria will predominate. To produce a cattle feed supplement from whey, it makes no differ— ence what the species of bacteria is as long as it produces predominate- ly lactic acid and is non-pathogenic for cattle. It is extremely unlike- ly that any pathogenic species could survive the restrictive conditions imposed on the whey fermentation. This conclusion is drawn from the fact that lactic acid fermentations are used as a "natural" means of preservation for such items as cheese, pickles, sauerkraut, and silage (63). The occurence of pathogenic bacteria is not a problem when the above items have been processed prOperly. 8.3 Effect of'pH in the Range of 5.5 to 6.0 on continuous Fermentation The effect of pH on the continuous fermentations of this study con- firms that the ammonium lactate system is affected by pH in a manner sim- ilar to the calcium lactate (37) and sodium lactate (20) systems. The 57 data indicate that, for a given flow rate, more lactose is converted at pH 6.0 than at pH 5.5. A logical conclusion would be to operate the sys- tem at pH 6.0. However, other species can also grow more efficiently at the higher pH. During most of these studies the pH was kept at 5.5 to discourage contamination while allowing the lactobacilli to ferment at a reasonable rate. The pH setting of 5.5 was selected because it was known that relative- ly few organisms tolerate this low pH at a temperature of hh C. However, some clostridia can ferment lactic acid to butyric acid even at these re— strictive conditions of pH and temperature (9, 37). There must be another factor which accounts for the fact that no butyric acid or spore-forming bacteria were observed in the product. To better understand the "other factor" which prevents contamination of the continuous whey fermentation, recall that a weak acid in an aque- ous medium is only partially dissociated. For example, + _ H-Lactate «(A ~4>- H + Lactate (l8) undissociated acid dissociated acid * Also recall that the presence of a salt of the lactate influences the con- centration of lactate ions in the system; and, in turn, the concentration of undissociated acid: H20 1\OH_ + _ H—Lactate «(———-)- H + Lactate (19) NH:- / NHE-Lactate 58 For a given pH (i.e., for a given concentration of H-ions), any addition of ammonium lactate to the system.must result in an increase in the con- centration of undissociated lactic acid. How much of the lactate from the added salt is converted to undissociated lactic acid will depend on the degree of dissociation of the salt at the given pH and temperature. It was demonstrated in 1928 that the concentration of undissociated lactic acid is more directly related to the inhibition of lactic acid bacteria than the actual pH of the medium.(55). More recently, a gener- alized model for the effect of pH on the biological activities of weak acids and bases was developed by Simon and Beevers (59). Both studies indicate that it is mainly the concentration of undissociated acid that inhibits the activity of a bacterial culture. For a given medium, there is a maximum concentration of undissociated lactic acid which a bacterial species can tolerate. The high concentration of undissociated lactic acid, therefore, is very likely the "other factor" which makes it possible to operate the con- tinuous whey fermentation at a pH above what would normally be considered necessary to exclude butyric acid-producing bacteria. Because of the high concentration of undissociated lactic acid in the product, it is quite likely that the continuous fermentation could be operated safely at a pH above 5.5 and thereby take advantage of the higher fermentation rates obtainable at the higher pH. During this study the process was operated at pH levels up to 6.0 without noticeable contamina- ‘tion. However, the process was not operated at the higher pH long enough 1K3 conclude definitely that no contamination would result. Because of iflie significantly better conversion of lactose to lactic acid at the high- er: pH levels, a very productive project would be the determination of the 59 maximum.pH which will still exclude undesirable microorganisms. The above discussion also leads to the conclusion that particular attention must be paid to pH when operating a multi-stage continuous whey fermentation. The concentration of ammonium lactate and hence the con- centration of undissociated lactic acid would be lower in the first stage than in the final stage. Therefore, a lower pH must be maintained in the first stage to maintain the same degree of inhibition obtained by a high- er pH in the final stage. The above discussion can also be related to batch fermentations. At the time of inoculation of a batch culture, the product concentration is low. Therefore, a relatively low pH would be required to obtain a given degree of inhibition of contaminants, and a pH of 5.5 has proven satis- factory for this purpose. The general practice has been to maintain the initial pH throughout the entire course of the fermentation. However, it should be possible to program.the pH so that it is raised as the fermen- tation progresses and still maintains adequate inhibition of contaminants. By raising the pH as the fermentation progresses, the product would be less inhibitory to the lactic acid bacteria while still inhibiting the less acid tolerant species. The net effect should be a significant re- duction in batch times. A recent publication by Hanson and Tsao (20) states that lactate concentrations in the range of 0 to 2% were not inhibitory to the lactic acid bacteria used in that study. The reduction in the specific growth rate for the bacteria during the progress of the fermentation was attrib- uted to the exhaustion of substrate. In view of the above discussion on product inhibition and in view of a study which used dialysis culture to prove product inhibition (16), it is difficult to accept the view that a 6O substrate concentration in excess of 10 g per liter was limiting the rate of fermentation. The fact that product inhibits lactic fermentations ac— counts for the extremely large values (> 20,000 mg per liter) reported for the saturation constant KS in the Monod equation. The value of KS is actually very low. For bacteria growing in car- bohydrate substrates, it lies in the order of magnitude of decades of milligrams per liter of medium (11). Lactic acid fermentations are nor- mally considered complete when the substrate level has been reduced to less than 1,000 mg per liter (56). Assuming a value of 10 mg per liter for KS, it can be demonstrated that the term S/(KS + S) is of little con- sequence even at the "end" of a lactic acid fermentation. s _ 1,000 _ 1,000 _ ~ K + s ' 10 + 1,000 ' 1,010 ' 0'990 ‘ 1 (20) Hence, KS in the study by Hanson and Tsao was more likely an indication of product inhibition rather than substrate exhaustion. Along with the fact that maintenance metabolism was ignored, this may explain why their mathematical model was not accurate when it was applied to continuous cul- ture. Finally, the concept of the interaction between the neutralized lactic acid and undissociated lactic acid helps to explain the difference between the inhibitory effect of ammonium lactate and calcium lactate. The amount of undissociated acid produced by the addition of the salt de— pends on the degree of dissociation of the salt. Ammonium lactate, which is a salt of weak acid and a weak base, will dissociate to a higher de— gree than calcium lactate, which is a salt of a weaker base. Therefore, it should be necessary to add more calcium lactate than ammonium.lactate to whey to obtain total inhibition of a bacterial species at a given pH. 61 In fact this happens. With reference to the horizontal line at pH 5.5 in Fig. 1h, it took h% lactate as ammonium.lactate to inhibit culture #2217 at pH 5.5. At the same time it took 8% lactate as calcium lactate to in- hibit culture #2217 at the same pH. If the concentration of undissociated lactic acid were determined, it should be about the same in both cultures if the above argument holds. 8.h Quality of'Product from continuous Fermentor The gas chromatograph is a very useful tool for evaluating the quali- ty of fermentation products. Besides an approximate analysis for lactic acid, it also analyzes for other short-chain metabolic products. The presence of other metabolic products besides lactic acid would be posi- tive indication of contaminating microorganisms (19). Fortunately, no products other than lactic acid plus traces of ethanol and acetate were found in the product from either the batch or continuous fermentor. Therefore, it is concluded that the lactic acid bacteria were not replaced by another group of bacteria even after operating the continuous fermentor non-aseptically for D2 days. Not only is it important that lactic acid is the sole product of fer- mentation, but it is also important that the product contain as low a con- centration of lactose as practical. Residual lactose is not desirable in the final product due to the fact that it crystallizes after the product is condensed. The lactose crystals precipitate in the product storage tanks to give a troublesome sludge, and they plug the product transfer lines. A reasonable goal would be a residual lactose concentration of less than 0.1% in the final product. This is easily achieved in vigorous batch fermentations, and it can be achieved in a continuous fermentation if more than one stage is used. 62 8.5 Effect of Yeast Emtract and cornsteep Liquor on Batch Fermentation The beneficial effect derived from yeast extract and cornsteep li- quor in the batch fermentations might seem surprising in view of the fact that milk is often referred to as "nature's most perfect food". Whey is simply milk from.which some protein and the butterfat have been extracted. Very few bacteria utilize lipids (e.g., butterfat) as nutrient sources, and whey retains a significant amount of protein (0.9%). Therefore, whey might be thought of as a "nearly perfect food" for bacteria. However, lactic acid bacteria are known to be extremely fastidious organisms (A7, 60, 6h). It is not unusual for lactic acid bacteria to be stimulated by various sources of growth factors even when the lactic acid bacteria are growing in complex media such as whey (7, 28, 53, 73). The results bore out this latter information. In a commercial operation to produce a high—protein cattle feed sup- plement from whey, it may be beneficial to supplement the whey with a source of growth factors such as cornsteep liquor. Arnott (3) recommended the use of a 10% inoculum to give a fermentation time of about 1h hr and to prevent spoilage. In approximately 50 pilot—scale batch fermentations conducted at Michigan State University (23) concurrently with this study, 2.5 to 5.0% inoculum was used along with 0.7% cornsteep liquor; and the fermentations were completed in approximately 1h hr. The use of 3 gal of sterile cornsteep liquor made it possible to save 25 to 38 gal of culture medium (skim milk). The cornsteep liquor cost approximately $1.20 and resulted in a savings of approximately $15.00 worth of skim milk. Additional cost savings could be expected from the ease of stor- ing the stable cornsteep liquor rather than the perishable milk. The cost-saving benefits from the use of growth factors would be less, however, 63 if whey were used as the inoculum medium or if a portion of a previous fermentation were used to inoculate a batch of whey. Although growth factors were not used in the present continuous fermentation studies, it is conceivable that the retention time required to achieve 98% lactose conversion could be reduced by the use of growth factors. Growth factors may be particularly beneficial in a multi-stage fermentation in which the first stages are not operating at high product concentration. Moreover, it has been demonstrated that lactobacilli are more acid tolerant in enriched media, and this phenomenon would likely benefit a supplemented continuous fermentation (55). In addition to yeast extract and cornsteep liquor, there are many other sources of growth factors which might prove equally useful in lac- tic acid fermentations. Other natural plant extracts have been used to stimulate fermentations: malt sprouts (7) and alfalfa solubles (33, 52) are two examples. A common practice is to use a mixed culture of a myco— derm to supply growth factors for the bacteria (73). In some instances, Maillard reaction products (28) and formic acid (66) stimulate lactic acid bacteria. Various pure compounds (e.g., riboflavin, pantothenic acid, nicotinic acid and biotin) are periodically used to supplement lac- tate fermentations (53, 60). Therefore, a wide variety of materials are available for evaluation as sources of growth factors in whey fermenta- tion. 8.6 Ammonium Lactate and Calcium Lactate Inhibition of Batch Fermentation The simple test used to study the effect of ammonium and calcium lac- tate on the fermentation of whey yields a considerable amount of informa- tion. First, the acid tolerance of a given strain of lactic acid bacte- ria can be evaluated with the test. This is easily done by inoculating 6h unsupplemented samples of whey with the culture in question and checking the final pH after 30 to ho hr of incubation. This test is useful for evaluating the acid tolerance of a strain, but it tells nothing about the rate at which the strain converts the substrate to lactic acid. Some lactic acid bacteria ferment the substrate very rapidly as long as the concentration of product is quite low, but the fermentation rate is considerably reduced as the product concentration increases. Strepto- coccus lactis is an example of a fast-fermenting but acid-sensitive species. Other species ferment the substrate more slowly but are less sensitive to the acid product (e.g., Lactobacillus bulgaricus). Similar variations occur in strains within a species. Generally it is desirable to have an acid tolerant species for con- tinuous fermentation because of the continually high concentration of product. For batch fermentations the acid tolerance of the species is not as critical if a relatively low concentration of sugar is fermented. As the initial concentration of sugar is increased, the ability of the culture to tolerate the product becomes more important. The relative acid tolerance of the two cultures which were tested is independent of the amount of product present. In Fig. 1h the two curves representing calcium lactate inhibition are nearly equidistant, and the same generally holds true for the two curves representing ammoni- um lactate inhibition. The top ammonium lactate line very likely pla— teaus because the initial pH was not high enough to permit bacterial growth (see Tables 5 — 8 in Appendix). The pH in the plateau region was reduced slightly from the initial value. This may have been the result of maintenance metabolism before death of the culture. The approximate amount of substrate which can be fermented by a 65 species under conditions of controlled pH can be estimated from Fig. 1h. This estimation can be made by drawing a horizontal line across the graph at the level at which the pH is to be controlled. At the point of inter- section between the horizontal line and the product inhibition curve, one can read the maximum amount of lactate which the culture can tolerate at that pH. For example, at pH 5.5 Culture #2217 could be expected to toler- ate a maximum lactate ion concentration of approximately h% when ammonia was the neutralizing agent and approximately 8% when calcium hydroxide was the neutralizing agent. These values are only approximations because some additional ammonium lactate (approximately 1%) was present in all the samples as a result of neutralizing the whey with ammonium hydroxide and because no accounting was made of the increase in volume caused by the addition of the lactate solutions. A more precise estimate of the maximum amount of fermentable substrate could be made by reconstructing Fig. 1h after analyzing each culture for the amount of lactate present. This analysis was not done since the original experiment was not de- signed to estimate the maximum fermentable substrate and because the dif- ference between the estimate given and the more precise value is not ex- pected to be very great. The above example indicates that, for a lactic fermentation neutral- ized with ammonia, the maximum amount of fermentable sugar is h to 5% which is somewhat lower than previous experience (5 to 7%). The differ- ence is likely due to the fact that the culture used in this particular experiment was inadvertently changed due to prolonged storage. In an earlier test with a culture of #2217 which had been maintained by re— peated transfers to fresh medium, the culture was able to tolerate a pH of 3.8 in an unsupplemented culture. 66 Finally, Fig. 1h illustrates that a given culture of lactic acid bacteria can ferment much more sugar if the acid product is neutralized with calcium hydroxide rather than ammonium hydroxide. This information is important to bear in mind when attempting to relate commercial lactic acid production (in which neutralization is accomplished with calcium hydroxide or carbonate) to the production of a cattle feed supplement from whey. The cattle feed serves as a source of nitrogen for the cattle. Therefore, ammonia is the neutralizing agent of choice. However, the use of ammonia to maintain a pH of 5.5 restricts the maximum level of fer- mentable substrate to a level (5 to 7%) considerably below the 10 to 20% used in commercial lactic acid fermentations. TABLE 5. 9. APPENDIX values for pH obtained during batch fermentations of'whey supplemented with various levels of’calcium lactate and inoculated with L. bulgaricus #2217 pH after Lactate ion, Z lactate addition inoculation 17 hr 41 hr 0 6.7 6.5 4.7 4.6 l 6.4 6.2 5.1 4.8 2 6.3 6.1 5.3 5.0 3 6.2 6.0 5.4 5.1 4 6.2 6.1 5.5 5.2 5 6.2 6.1 5.5 5.2 6 6.1 6.1 5.6 5.3 7 6.2 6.0 5.6 5.4 8 6.2 6.1 5.7 5.5 9 6.2 6.0 5.7a 5.5at 10 6.2 6.1 - a - a a Very viscous slurry TABLE 6. Values for pH obtained during batch fermentations of whey supplemented with various levels of’calcium lactate and inoculated with a mixed culture of'lactic acid bacteria pH after Lactate ion, 2 lactate addition inoculation 17 hr 41 hr 0 6.6 6.3 4.4 3 9 1 _ - _ - 2 - l - 3 a - _ - 4 6.1 6.0 4.9 4.2 5 6.1 6.0 5.0 4.3 6 6.1 5.9 5.1 4.4 7 6.1 6.0 5.1 4.5 8 6.1 6.0 5.1 4.5 9 6.2 6.0 5.13 4.68 10 6.2 5.9 5.1 4.7 aVery viscous slurry 67 68 TABLE 7. Values for pH obtained during batch fermentations of'whey supplemented with various levels of’ammonium lactate and inoculated with a mixed culture of’lactic acid bacteria pH after Lactate ion, 2 lactate addition inoculation 17 hr 41 hr 0 6.7 6.4 4.2 3.8 1 6.8 6.5 4.5 4.1 2 6.9 6.4 4.8 4.3 3 6.9 6.4 4.9 4.4 4 6.7 6.3 5.2 4.5 5 6.6 6.2 5.4 4.7 6 6.5 6.1 5.6 4.8 7 6.4 6.1 5.7 5.2 8 6.3 6.0 5.8 5.4 9 6.2 5.9 5.6 5.5 10 6.2 5.9 5.7 5.5 TABLE 8. Values for pH obtained during batch fermentations of'whey supplemented with various levels of’ammonium lactate and inoculated with L. bulgaricus #2217 pH after Lactate ion, Z lactate addition inoculation 17 hr 41 hr 0 6.6 6.6 4.7 4.6 1 6.6 6.7 5.1 4.9 2 6.7 6.7 5.3 5.1 3 6.6 6.6 5.6 5.3 4 6.6 6.6 5.8 5.5 5 6.5 6.5 5.9 5.7 6 6.4 6.3 6.0 5.8 7 6.2 6.3 6.0 5.8 8 6.2 6.2 6.0 5.8 9 6.1 6.1 5.9 5.7 10 6.1 6.1 5.9 5.7 10. ll. 12. 10. BIBLIOGRAPHY Allen, C. K., H. E. Henderson, and W. G. Bergen. 1971. Ammonium salts as a source of crude protein for feedlot cattle, p. l6-27. In Report of beef cattle research 1971A. Michigan State University. AH-BC-703. Anderson, R. F. 1970. Whey problems of the cheese industry, p. 24- 29. In Proceedings-Whey Utilization Conference. ARS-73-69, U.S. Department of Agriculture, Philadelphia, Pa. Arnott, D. R., S. Patton, and E. M. Kesler. 1958. A method for manufacturing a high-nitrogen - lowhlactose product from whey. J. Dairy Sci. 41:931-941. Beaman, R. G. 1967. Vinegar fermentation, p. 344—359. In H. J. Peppler, Microbial technology. Reinhold Publishing Corp., New York. Bellamy, W- D. 1973. Conversion of insoluble agriculture wastes to SCP by thermophilic microorganisms. Report given at International Conference on SCP at M.I.T., Cambridge, Mass. (General Electric Co., Schenectady, New York). Beran, K. 1966. Continuous cultivation in applied microbiology, p. 386. In I. Malek and Z. Fencl (ed.), Theoretical and methodolog- ical basis of continuous culture of microorganisms. Academic Press, Inc., New York. Campbell, L. A. 1953. Production of calcium lactate and lactic acid from cheese whey. Can. Dairy Ice Cream J. Marchz29-31; 77-78. Carlsson, Jan. 1973. Simplified gas chromatographic procedure for identification of bacterial metabolic products. Appl. Microbiol. 25:287-289. Casida, L. E., Jr. 1968. Anaerobic fermentations, p. 304-314. In L. E. Casida, Jr., Industrial microbiology. John Wiley and Sons, Inc., New York. Czarnetzky, E. J. 1959. Treating whey. U.S. Patent No. 2,904,437. Fencl, Z. 1966. Theoretical analysis of continuous culture system, p. 69-153. In I. Malek and Z. Fencl (ed.), Theoretical and methodolog- ical basis of continuous culture of microorganisms. Academic Press, Inc., New York. Finn, R. K., H. 0. Halvorson, and E. L. Piret. 1950. Lactic acid fermentation rate. Ind. Eng. Chem. 42:1857-1861. 69 l3. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 70 Fisher, C. H., M. B. Dixon, and E. M. Filachrone. 1951. Bibliog- raphy on lactic acid and derivatives. U.S. Department of Agriculture, Fisher, C. H., and E. M. Filachione. 1950. Properties and reactions of lactic acid — a review. U.S. Department of Agriculture, AIC-279.~ Foster, E. M., F. E. Nelson, M. L. Speck, R. N. Doetsch, J. C. Olson, Jr. 1957. Dairy microbiology. Prentice-Hall, Inc., Englewood Cliffs, N. J. Friedman, M. R., and E. L. Gaden, Jr. 1970. Growth and acid produc- tion by Lactobacillus (L.) delbrueckii in a dialysis culture system. Biotech. Bioeng. 12:961-974. Gallup, D. M. 1962. Concentrated culture of microorganisms in dialysis flask and fermentor systems. Ph.D. Thesis, Department of Microbiology, University of Michigan, Ann Arbor. Gerhardt, P., and M. C. Bartlett. 1959. Continuous industrial fer- mentations. Adv. Appl. Microbiol. 1:215-260. Haldeman, L. V., and w. E. c. Moore (ed.). 1973. Anaerobe labora- tory manual, 2. Anaerobe laboratory, Virginia Polytechnic Institute, Blacksburg. Hanson, T. P., and G. T. Tsao. 1972. Kinetic studies of the lactic acid fermentation in batch and continuous culture. Biotech. Bioeng. 14:233-252. Hazzard, D. G., E. M. Kesler, D. R. Arnott, and S. Patton. 1958. Use of high-nitrogen feedstuff made from whey in the rations of dairy animals. J. Dairy Sci. 41:1439-1445. Henderson, H. E. 1973. New liquid feed technology. Paper presented to National Fertilizer Solutions Association Convention, New Orleans. (Michigan State University, East Lansing). Henderson, H. E., and C. A. Reddy. 1973. A method for the commer- cial production of BACTOLAC from acid whey. Progress report for period ending 11 May 1973. Michigan State University, East Lansing Herbert, D. 1961. A theoretical analysis of continuous culture systems, p. 21-53. In Continuous culture of microorganisms, S.C.I. Menograph No. 12. Society of Chemical Industry, London. Hodge, H. M., and F. M. Hildebrandt. 1954. Alcoholic fermentation of molasses, p. 73-94. In L. A. Underkofler and R. J. Hickey, Indus- trial fermentations. Chemical Publishing Co., Inc., New York. Horwitz, W. (ed.). 1970. Official Methods of Analysis of the Asso- ciation of Official Agricultural Chemists, 11th ed. Association of Official Agricultural Chemists, Washington, D. C. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39- 40. 71 Jansen, H. C. 1945. Method for preparation of alkali lactates, especially ammonium lactate through lactic acid fermentation of sugar containing solutions. Dutch Patent No. 57.848. Jemmali, M. 1969. Influence of the Maillard reaction products on some bacteria of the intestinal flora. J. Appl. Bact. 32:151—155. Johnson, M. J. 1941. Isolation and prOperties of a pure yeast poly- peptidase. J. Biol. Chem. 137:575. Jones, R. W. 1951. (A general graphical analysis of continuous reac- tions in series of agitated vessels. Chem. Eng. Prog. 47:46-48. Kempe, L. L., H. 0. Halvorson, and E. L. Piret. 1950. Effect of continuously controlled pH on lactic acid fermentation. Ind. Eng. Chem. 42:1852-1857. Kempton, A. G., and C. L. San Clemente. 1959. Chemistry and micro- biology of forage-crop silage. Appl. Microbiol. 7:362-367. Kohler, G. 0., and E. M. Bickoff. 1971. Commercial production from alfalfa in USA, p. 69-77. In N. w. Pirie (ed.), Leaf protein: its agronomy, preparation, quality and use. Blackwell Scientific Publi- cations, Oxford. Longsworth, L. G., and D. A. MacInnes. 1935. Bacterial growth with automatic pH control. J. Bacteriol. 29:602-607. Longsworth, L. G., and D. A. MacInnes. 1936. Bacterial growth with at constant pH. J. Bacteriol. 31:287-300. Longsworth, L. G., and D. A. MacInnes. 1937. Bacterial growth at constant pH. J. Bacteriol. 32:567—585. Luedeking, R. 1956.' The lactic acid fermentation at controlled pH: kinetics of the batch process and continuous flow theory and experi- ments. Ph.D. Thesis, Department of Chemical Engineering, University of Minnesota, Minneapolis. Luedeking, R., and E. L. Piret. 1959. A kinetic study of the lactic acid fermentation: batch process at controlled pH. J. Biochem. Microbiol. Technol. Eng. 1:393-412. Luedeking, R., and E. L. Piret. 1959. Transient and steady states in continuous fermentation: theory and experiment. J. Biochem. Microbiol. Technol. Eng. 1:431—459. Malek, I. 1966. The role of continuous processes and their study in the present development of science and production, p. 11-30. In I. Malek and Z. Fencl (ed.), Theoretical and.methodological basis of continuous culture of microorganisms. Academic Press, Inc., New York. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 5).. 55. 72 Marth, E. H. 1970. Fermentation products from.whey, p. 43-81. In .B. H. Webb and E. 0. Whittier (ed.), By-products from.mi1k. Avi Publishing Co., Westport, Conn. Maxon, W. D. 1955. Continuous fermentation: a discussion of its principals and applications. Appl. Microbiol. 3:110-122. Maxon, W. D. 1960. Continuous fermentation. Adv. Appl. Microbiol. 2:349-355. Maxon, W. D. 1960. Continuous fermentation. Ind. Eng. Chem. McCullough, M. E., W. E. Neville, Jr., and W. J. Menson. 1972. Ammoniated whey as an ingredient in complete livestock rations. Feedstuffs. 18 December:27, 54- Novick, A. 1955. Growth of bacteria. Ann. Rev. Microbiol. 9:97-110. Orla-Jensen, S. 1931 Dairy bacteriology. 2nd ed. (in English). P. Blakiston's Son and Co., Inc., Philadelphia. Peppler, H. J. 1967. Ethyl alcohol, lactic acid, acetone-butyl alcohol and other microbial products, p. 403-416. In Harry J. Peppler (ed), Microbial technology. Reinhold Publishing Corp., New York. Peppler, H. J. 1967. Yeast technology, p. 145-171. In H. J. Peppler (ed.), Microbial technology. Reinhold Publishing Corp., New York. Perquin, L. H. C. 1946. Method for the preparation of lactic acid, lactic acid compounds and ammonium lactate containing cattle feed. Dutch Patent No. 58,545. Perry, N. A., and F. V. Doan. 1950. A picric acid method for the simultaneous determination of lactose and sucrose in dairy products. Pirie, N. W. 1971. The use of the by-products from leaf protein ex- traction, p. 135-137. In N. W. Pirie (ed.), Leaf Protein: its agron— omy, preparation, quality and use. Blackwell Scientific Publications, Oxford. Prescott, S. C., and C. G. Dunn. 1949. The production of lactic acid by fermentation, p. 404-428. In S. C. Prescott and G. G. Dunn, Industrial microbiology. McGraw-Hill, New York. Ricica, J. 1958. Continuous culture techniques, p. 75—105. In Continuous cultivation of microorganisms - a symposium. Czechoslovak Academy of Science, Prague. Rogers, L. A., and E. 0. Whittier. 1928. Limiting factors in the lactic fermentation. J. Bacteriol. 16:211-229. 56. 57. 58. 59- 60. 61. 62. 63. 64. 65. 66. 67. 68. 690 73 Schopmeyer, H. H. 1954. Lactic Acid, p. 391-419. In L. A. Underkof- ler and R. J. Hickey (ed.), Industrial fermentations. Chemical Pub- lishing Co., Inc., New York. Schultz, J. S., and P. Gerhardt. 1969. Dialysis culture of micro- organisms: design, theory, and results. Bacteriol. Rev. 33:1-47. Silver, R. S., and P. G. Cooper. 1972. Production of single cell protein (SCP) from normal paraffins. Report given at American Chemi- cal Society, New York meeting. (Gulf Research and Development Co., Pittsburgh, Pa.) Simon, E. W., and H. Beevers. 1952. The effect of pH on the biologi- cal activity of weak acids and bases. New Phytologist 51:163-189. Snell, E. S. 1952. The nutrition of the lactic acid bacteria. Bacteriol. Rev. 16:235-241. Solomons, G. L. 1972. Improvements in the design and operation of the chemostat. J. Appl. Chem. Biotechnol. 22:217-228. Sortland, L. D., and C. R.'Wilke. 1969. Growth of Streptococcus faecalus in dense culture. Biotech. Bioeng. 11:805-841. Stanier, R. Y., M. Doudoroff, and E. A. Adelberg. 1970. The lactic acid bacteria, p. 663-672. In R. Y. Stanier, M. Doudoroff, and E. A. Adelberg, The microbial world. Prentice-Hall Inc., Englewood Cliffs, N. J. Stiles, H. R., and L. M. Pruess. 1938. Nutrient requirements of L. delbrueckii in the lactic acid fermentation of molasses. J. Bacteriol. 36:149-153. Vaughn, R. H. 1954. Acetic acid - vinegar, p. 498-535. In L. A. Underkofler and R. J. Hickey, Industrial fermentations. Chemical Publishing Co., Inc., New York. Veringa, H. A., T. E. Galesloot, and H. Dovelaar. 1968. Symbiosis in yoghurt (II): isolation and identification of a growth factor for Lactobacillus bulgaricus produced by Streptococcus thermophilus. Neth. Milk Dairy J. 22:114-120. Ward, K. W. 1970. Industrial processing of whey today, p. 30-35. In Proceedings - Whey Utilization Conference. ARS—73—69, U.S. Depart- ment of Agriculture, Philadelphia, Pa. Webb, B. H. 1970. Utilization of whey in foods and feeds, p. 102— 111. In Proceedings - Whey Utilization Conference. ARS-73-69, U.S. Department of Agriculture, Philadelphia, Pa. Weber, A. P. 1953. The design of commercial continuous reactor sys- tems. Chem. Eng. Prog. 49:26. 70. 71. 72. 73. 71. Whey Products Conference Proceedings. 1972. (ERRL Publ. N0. 3779). Agricultural Research Service, U.S. Department of Agriculture, Phila- delphia, Pa. Whey Utilization Conference Proceedings. 1970. ARS-73—69, Agricul- tural Research Service, U.S. Department of Agriculture, Philadelphia, Pa. Whittier, E. 0., and L. A. Rogers. 1931. Continuous fermentation in the production of lactic acid. Ind. Eng. Chem. 23:532-534. Whittier, E. 0., and B. H. Webb. 1950. By—products from milk. Reinhold Publishing Corp., New York. Ill. ~‘ ‘ ... u HICHIGQN STATE UNIV. LIBRQRIES II" III III 1 WI II III”! llllll 31293 07000824