II III III IIIII‘III; I I . 111M} {J}! I‘n1ll'r ‘ . y amt)?!” ) I “I130 : ””11; ”III'IIIII1IJ:IJ1JJ..I¢ 1713' IIIEI‘I I‘IIIIHI , ”1- I I yIIIIwI' ”TM; t ,I. . 'LV‘QHEI “Lt: tI “1".“ '1!“ H . ' - 2 J I , H‘ ' ' 1/ thf 4‘1"" in. .1! 12‘, -« if" V" II' “ “4m; V 'r» N"! . ' :QI"1Pr1r(hn‘f' H ‘1‘ IIIIIIl ‘5' IIIIEI 111;” In W" III III I II IIIII IIIIIIIIIIII ;«I y‘ ..rI I IIIIaIIIIIIII II III _II‘ KLII mm“ 11"?“ ,1: VI ‘ .1”; ‘IA‘ ' ;\ ‘51.. 1411:“: 2, :flii fig:";"'f“'l'!‘l"‘ 3.le ‘4 (119311” N ;,_ ”M "' S‘J‘ '55». !' J ’“J " W I: "s’i‘.’ 4“" ' 1ft“ 11m (59}: ' (I If IIJIII'IIhH FlI LL 2.; ”3.215;... ..~A- -4... ’$ fin—é‘Z.‘ 7‘14 I.“ *1. 1’31 «II ”2*"; 4h 52;? I Q :11 :11: 4-2: ._;:.. 7,3, “1:“ ‘ v o ‘7 c: .. ......._ 12%“?— w- -=: *5sz ' "-7; . W” .—t‘”% 25*: «flair. A . --.g§;” , %_f; ..:_' I A 9; 1:3; ‘3 , as, - m...._‘_?£:_‘:— . . A - .0.— $3.. .. »-—.~ ’3‘}. a. u tau-452:“:21 - ,— 127':- “44,“. .‘ 7—235 mate. 1 ":— :‘attfic': . . ’1‘.“ 75%: < .- 735: r’". -‘- .. . , ‘1 :5"; 4... 122:113‘-"""“' 7731:};— 'EE% tJ§§§sé- _.-__ 3;:fiff-x ' ."‘:‘:: ‘1. ;-:T._. 4 ~qa=m .v- alga 2'3; 2— ‘73:" u. .. .1" mm. tat—"firs:— < ”a. "a: ~“gg‘;駧7¥i.i -S‘ -133: “C“ .y? we“ 7 t‘FV .«g. *H :" a. is: “43; b IIIIIII .Io- . mg“ ' Hm I J $- ' 'LF‘; ' h '31} rm]. '» ’} #195313?“ hgik‘b I ,. -R‘ my); a ‘ MKISEEW‘1'3M "1 7. '1. a. . _.._.. h ‘ J :J .4 Mm" -fimv - “’3‘: f". _....,_-.:.. . .. ;SETL.-. ‘ < 7““ ~.":; ”:5: j :4 m” —v-G.~"‘—"”:’i1 yam—”h": .‘ - 313:. JC’; 2—. . ‘57.": ~: «first: ‘ _r v. 7'" . . 4.55:», ”z. .. ;:=.. .92.»; « ‘ttafi’ " ”:553'11' #5.?“ cr—w £4“... 1. 1; .J. ‘1 urfi. #1191135“? :3 ‘1' III IIIm MW.“ J’J’IJII 1:1 “REL: 110 .w—vw— .. £51» A h A 1. .7- “is“ .252? M. W.-.” .- 72":in _ “-51% E I‘xfl.“': ' : m 7:3? .4 r 5%.] I 33 In} if! i“ W 1:332. ~*::;:*-' ' - 3:311... 7:47 .3.“ .. a; -—- I! -:§gf¥v .._ "7:13. é J7--. ,:.:, : .:_ -..._. ‘. ml . A 3" “if?" .3 f >% 4...: . “J 211%.”:9 .. ‘3 . “2.... .. 77%., 3‘ :fflv '7' “491—.” w-.- (T; :u'_". 7‘4- ‘ r ”.4 13‘7““? «T‘J v16" 1 n) A“ f4 -1 11. , g5; '. {$43311 1&va i ‘1 (It: ll‘ , I) 341-114“ m IIII: Ir . I dun {WI I128]: 451%“! I I-III'I‘ IIII "£11113? 4g II‘TI' I 13-3 134 ' 7 ~, r553“ ' 37...- .. ~ m2”: . Z 23 .‘7 . I: . . _ y: .1. u." 7&3’3: . ., ‘Q‘? . _‘ I: .-~A - “ '5: -' ”T. 1 . A %_ LE4]: o :5 -.-r- . ‘I . A 4-. ' _ ‘ ' . ‘ ” ” ”a V' .J’iz'té-rvq-A % "DJ“ ,_ ,. v! ‘— Ll“ 3'11, 131% #5}? #11311 IIIEIII KI)“: ’; :1“& ”My: 9% "fifi - {Jh‘ !J& “J? JJJ ‘21:: $33.24; 54% .9‘: 7:4. » . W 43"”; It" 435:} ~' g F {éiixna’o- 3 1 222:: H v" :4- 3.7:: ‘7 fl...) («3:1 ‘ :7? '.-‘:I . '23-; V “in? 2' - a _ _._:... . ’7’» >. no. cngfizfl; " w. .4 ._ “7:. .8“ 4,: ~-:- .3. .Jr' it? :1 r: '3 -m- 2.1“.“ 2):. .53. #1:. " :7 . ‘2' ‘ 5 “:E‘j .‘H 3“ .III‘q‘I' I J :I ‘ ZIIIII‘OJ ‘I . Jigs}: , -m§1r v fisJ' ~23: ‘ (1 7 I": {w ’ {1‘1 J J. 4, ., r .. 5‘5”“ : . - l '1: 1224M h;‘§‘fi4’ . 7 “(my - 3% ‘5‘}: . '43“: "3‘3.” ~ 1‘": :V.)‘. i 0 JV»; 4 q I n- - ww'fiitV-SPE J .J H ~ $194” -. ‘ THEE“??? This is to certify that the thesis entitled MICROFILTRATION PROCESS FOR ENHANCED PRODUCTION OF Y‘DNA RECEPTOR CELLS OF ESCHERICHIA COLI presented by KEVIN WARREN ANDERSON has been accepted towards fulfillment of the requirements for M. S . degree in Chemical ” Engineering 6‘“;in Major professor Date W3 0-7639 MS U is an Affirmative Action/Equal Opportunily Institution MSU LlBRARlES .——. V RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. MICROFILTRATION PROCESS FOR ENHANCED PRODUCTION OF rDNA RECEPTOR CELLS OF ESCHERICHIA COLI BY Kevin Warren Anderson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemical Engineering 1983 Copyright by KEVIN WARREN ANDERSON 1984 ABSTRACT MICROFILTRATION PROCESS FOR ENHANCED PRODUCTION OF rDNA RECEPTOR CELLS OF ESCHERICHIA COLI By Kevin Warren Anderson Escherichia coli HBlOl was cultured in a micro- filtration fermentor system to five times the density obtained in a control batch system. A synthetic glucose- salts medium was fed continuously into the aerated culture, and spent medium and products were continuously removed at the same rate through the microfiltration unit so that culture volume remained constant. The influence of hydraulic residence time and feed composition on the unsteady state growth of cells, consump- tion of nutrients and production of acids was studied. The culture exhibited changes in aerobic metabolism from aerobic fermentation to respiration depending on the glucose concentration in the culture. When glucose was in excess, aerobic fermentation occurred and mixed acids ac- cumulated. When glucose was limiting or exhausted, a res- piratory metabolism predominated with low acid production rates. ACKNOWLEDGMENTS I wish to express my appreciation to Dr. Eric A. Grulke for his help and guidance throughout my graduate studies and research and to Dr. Philipp Gerhardt from the Department of Microbiology for his guidance and use of his facilities. I wish to thank Drexelbrook Engineering Co. for their donation of a liquid level transmitter for use on this project and Mr. Dale Hauk of Gelman Sciences for assistance in the use of their products. I am further grateful to Dr. C. A. Reddy of the Department of Microbiology for use of equipment in his lab and Sue Cuppett of the Department of Food Science for advise on gas chromatography. I also acknowledge advise given by Dr. Donald K. Anderson and Dr. K. Jayaraman, both of the Department of Chemical Engineering, on process control applied in this investigation for which I am grateful. I acknowledge financial support from grant DAR 79—10236 (to Dr. Philipp Gerhardt) from the U. S. National Science Foundation. iii TABLE OF CONTENTS List of Tables . . . . . . . . . List of Figures . . . . . . . . Nomenclature . . . . . . . . . . 3.1 3.2 Nomenclature used in Literature General Nomenclature . . . Introduction . . . . . . . . . . Literature Review . . . . . . . 5.1 UIU1 5.4 Kinetics of Microbial Growth Review 5.1.1 Lag Phase . . . . . . . . . 5.1.2 Exponential Growth Phase . . 5.1.2.1 Unstructured Models 5.1.2.2 Structured Models . 5.1.3 Stationary Phase . . . . . 5.1.4 Death Phase . . . . . . . . Growth Kinetics of Escherichia COZi Membrane Enhancement of Microbial Cultivation . . . 5.3.1 Dialysis . . . 5.3.2 Microfiltration Liquid Level Control . . Materials and Methods . . . . 6.1 6.2 Organism and Culture Conditions Analytical Methods . . . 6.2.1 Optical Density . 6.2.2 Viable Cell Counts 6.2.3 Cell Dry Weight . 6.2.4 Ammonia . . . . . . 6.2.5 Glucose . . . . . . 6.2.6 Volatile Fatty Acids 6.2.7 Non-Volatile Acids . Experimental Microfiltration System and Procedure . . . . . . Liquid Level Control . . 6.4.1 Control Loop Analysis iv vii ix 7. 10. 11. 12. Results and Discussion . . . . . . . . . . . . . 7.1 Experimental Results and Discussion . . . . 7.1.1 Control Batch Cultivation: Experiment 1 . . . . . . . . . . . . 7.1.2 Microfiltration Cultivation . . . . 7.1.2.1 7.1.2.2 7.1.2.3 Long Residence Time Micro- filtration: Experiment 2 . Moderate Residence Time Microfiltration: Experi- ment 3 . . . . . . . . . . Short Residence Time Micro- filtration: Experiment 4 7.2 Mathematical Model and Computer Simulation . 7.2.1 Mathematical Model . . . . . . . . . 7.2.2 Kinetic Parameter Estimation . . . 7.2.3 Computer Simulation . . . . . . . 7.2.3.1 Batch Growth Simulation . . 7.2.3.2 Experiment 2 Simulation . . 7.2.3.3 Experiment 3 Simulation . . 7.2.3.4 Experiment 4 Simulation . Conclusion . . . . . . . . . . . . . . . 8.1 Equipment Considerations . . . . . . . . 8.2 Metabolic Considerations . . . . . . . . . 8.3 Cell Productivity Considerations . . . . . 8.4 Modeling Considerations . . . . . . . . . . Recommendations . . . . . . . . . . . Appendices . . . . . . . . . . . . . Appendix I: Standard Data . . . . . . Appendix II: Appendix III: Appendix IV: Sample Calculations . . . Numerical Data . . . . . . Level Control LOOp Start-Up Procedures and Schematic Diagram . . . Appendix V: Computer Program and Output . Literature Cited Materials Bibliography . . . . . . . . . . . . - 105 . 52 . 60 O 62 . 7O 76 . 82 . 82 85 9O . 91 96 . 111 . 111 . 112 .113 .113 .115 .116 .116 .123 .126 .131 .134 .141 .146 Table A1 A2 A3 A4 A5 A6 LIST OF TABLES Non-Volatile Acid Identification . . . Summary of Microfiltration Exper— iments . . . . . . . . . . . . . . . Early Exponential Growth Phase Specific Growth Rates (hr . . . . . . . . . Kinetic Parameters for the Mathematical Model . . . . . . . . . . . . . . . . Computation of Non-Volatile Acid Concentrations . . . . . . . . . . Numerical Data for Experiment 1 . . . . Numerical Data for Experiment 2 . . . . Numerical Data for Experiment 3 . . . Numerical Data for Experiment 4 . . . . Computer Program and Simulated Results vi 86 90 125 127 128 129 130 134 Figure 10 ll 12 l3 14 15 16 17 LIST OF FIGURES Comparison of the Dialysis Culture System and the Microfiltration System . . . . Cell Dry Weight vs. Optical Density at 600 nm (O.D.600) Standard Curve . . . . . Ammonium Ion Absorption Spectrum . . . . . Volatile Fatty Acid Chromatograms . . . . . Non-Volatile Acid Chromatograms . . . . . . Experimental Process Flow Diagram and Photograph . . . . . . . . . . . . . . Photograph of the Acroflux Capsule . . . Level Control LOOp Diagram . . . . . . . . Level Control Block Diagram . . Pump Speed (R) vs. Controller Output Current (ID) . . . . . . . . . . . . . Results of Experiment 1 . . . Proposed Modification of the TCA Cycle to a Branched Pathway for Exponentially Growing E. coli Under Aerobiosis (adapted from Amarsingham and Davis, 1965) . . . Cut—Away View of the Modified Gelman Acroflux Capsule . . . . Results of Experiment 2 . . . . . . . . Metabolic Shifts in E. coli under Aerobic Conditions . . . . . . . . . . . . . . . . Results of Experiment 3 . . . . . . . . . Results of Experiment 4 . . . . . . . . vii 26 29 33 37 41 44 46 48 50 54 58 61 65 69 72 78 Figure Page 18 Cell Dry Weight (xf) vs. Viable Cells (Nf) Concentrations from Experiment 1 . . . . 87 19 Computer Simulation of Experiment 1 . . . . . 93 20 Computer Simulation of Experiment 2 . . . . . 98 21 Computer Simulation of Experiment 3 . . . . . 102 22 Computer Simulation of Experiment 4 . . . . . 107 A1 Glucose Standard Curves . . . . . . . . . . . 117 A2 Ammonium Ion Standard Curves . . . . . . . . 119 A3 Standard Curves for Acetic Acid . . . . . . . 120 A4 Standard Curve for PrOpionic Acid . . . . . . 121 A5 Non-Volatile Acid Standard Curves for Experiment 4 . . . . . . . . . . . . . . . . 122 A6 Level Control LOOp Schematic Diagram . . . . 132 viii NOMENCLATURE 3.1 Nomenclature used in Literature Review < 71H velocity of respiration ix Symbol Units a maintenance metabolism parameter hr:i _2 A constant gl hr_l ATP concentration of adenosine tri- mole 1 phosphate b quantitative index of the Pasteur dimensionless effect —1 B constant hr_l —1 C constant in equation 11 gl hr_l C constant mole g C percent saturation of O dimensi nless L 2 -? Cl constant mole g -1 C2 constant mole mole_l Ed activation energy for death kcal_Tole Fd filtrate flow rate 1 hr_l Ff feed rate to fermentor l hr_l F purge rate 1 hr_l FE feed rate to reservior 1 hi -kd specific death rate hr_l -kd standard specific death rate hr -1 Ki° inhibition constant 9 l KL oxygen saturation constant dimegsionless KS substrate saturation constant 9 l_l Kl product inhibition constant 1 g n exponent in equation 8 dimensigpless N viable cell number cells 1 No initial viable cell number 9 l_l P product concentration g l_l Pmax maximum product concentration 9 1 allowing growth -1 -1 r rate of growth 9 l-lhr-l rg rate of product formation 9 l_lhr_l -r§ rate of substrate utilization g 1 hr -1 —l R universal gas constant k cal mole K S substrate concentration g 1 t time hr t lag time hr iag temperature K Vf fermentor volume 1 -l VG velocity of glycolysis hr_l V maximum velocity of glycolysis hr 9M reservoir volume 1 maximum velocity of respiration cell dry mass concentration initial cell dry mass concentration cell growth yield Greek Letters product yield product formation by maintenance metabolism specific growth rate maximum specific growth rate Other Subscripts '1» Wild-h 0.: Hal-ho For—4063636) HIM Dram Bran rmrh L-immx Hmo ) set 223 glycolytic quantity respiratory quantity 3.2 General Nomenclature ammonium ion concentration in feed ammonium ion concentration cross sectional area of fermentor (equation 23) deviation of filtrate flow rate from steady state rate nutrient feed rate deviation of feed rate from steady state rate controller transfer function filter transfer function motor and pump transfer function process transfer function controller output current deviation of controller output cur- rent from steady state current proportional gain substrate saturation constant transmitter gain deviation in fermentor liquid level from steady state deviation in fermentor liquid level from set point single cell mass viable cell count as colony forming units cells ml.-1 Symbol DUWUHHHH'UW mfijmiwtfirh EDD” acetic acid concentration in fermentor acetic acid concentration in feed rate of ammonium ion utilization rate of growth rate of product formation rate of glucose utilization pump speed deviation in pump speed from steady state speed Laplace transform independent variable residual glucose concentration in fermentor glucose concentration in the feed time cell dry weight concentration Greek Letters ammonium ion consumption coefficient glucose consumption coefficient maintenance metabolism parameter acetic acid yield coefficient controller derivative action controller integral action hydraulic residence time maximum specific growth rate xi INTRODUCTION The large scale production and recovery of new products made available by recombinant DNA technology will create new problems in process design and manufacturing. Desired products may be retained in the cell (intracellular) or released to the environment (extracellular). The new genetic information may handicap the receptor organism, and wild type revertants would have a selective advantage. Nutrient, temperature, and pH requirements of the organism must be satisfied in order to provide a suitable environment for growth. All of these factors are important when considering the production and recovery of these biological products. The usual method for production of biological products consists of a batch fermentation in large vessels with sub- sequent recovery steps to purify the desired products. If higher density populations of cells within the fermentor could be obtained, then smaller scale equipment would be re- quired or an existing process could be improved. Growth is typically limited in the batch fermentor either by depletion of a particular nutrient and/or the accumulation of toxic:end products> KS, equation (5) becomes = _ umax 6 P 1 + KS/S + S/KI’ ( ) for inhibitory substrates. 8 Aiba et al. (1968) and Levenspiel (1980) prOposed other forms to include product inhibition. The definitions of the symbols are given in the literature review. -K1P Aiba, et aZ.: u = “max (S+K ) (7) S Levenspiel : u = “max (§§R") (1 - g )n (8) S These models require knowledge of the substrate and product concentrations as functions of time and hence are models for the intrinsic rates of substrate utilization and product formation. Monod (1949) found that the total yield of cells in a batch culture was linearly related to the amount of limiting nutrient consumed and was independent of the concentration of nutrient. If the yield is also independent of the growth rate, then the rate of substrate utilization, -rS (the rate of formation of any component is taken as +), is given by I H II iHso pumpcmum Aooo.a.ov E: 000 um muflmcwa Hmoflumo .m> ucmflwz mun HHOU .m wusmflm 890.0 N o m c n N p o IJ. u 41 . . d O O .. 0 he... a 1 . ql w r . l °.P A . . .1 u .. M PwooundumaOSm. H. 9 I O i m; H .1 my / r AMAHN “U 0 N _ . p _ b b l m.“ 27 To determine the extent of interference caused by ions and organic components in the media, three solutions were pre- pared. The first contained 2.0000 g/l NH4C1 in water; the 4)2 SO4 in H20 and the third contained 2.0000 g/l NH4C1 in a solution with the same com— second contained 2.4706 g/l (NH position as the fermentation media. Thus, each solution con- + 4 0 water to give a final ammonium ion concentration of 16.16 tained 0.6735 g/l NH Each solution was then diluted with and 5.39 g/ml. Each was nesslerized along with water blanks as described below and the absorption spectrum was recorded between 380 and 500 nm with a grating spectrophotometer (Figure 3). There was a slight difference in the spectra near 400 nm for the solutions containing NH4C1 and (NH4)2 SO4 prepared in water probably due to dilution error. How- ever, the solution prepared in the fermentaton media show significant variation over the entire spectrum. Samples were typically diluted 1:100 with water such that the ammonium ion concentration was 1.0-15.0 ug/ml. To min- imize the error due to background interference caused by other ions present in the samples, a diluent was prepared with a composition 1/100 of the fermentation medium without NH4C1. A stock solution was prepared in the diluent with 100 ug/ml NH4C1 (33.67 ug/ml NH4+). Standards were prepared by pipeting 1.0, 2.0, 3.0, 4.0, and 5.0 ml of stock solution into testtubes and adding sufficient diluent to bring the volume to 10 m1. One ml of each standard and diluted sample was placed in- to test tubes. Two ml of Nessler's reagent and 3.0 m1 of Figure 3. 28 Ammonium Ion Absorption Spectrum. Three solutions were prepared to determine the interference by chemical species present in the media. They contained NH4C1 in water (0), (NH4)2 SO4 in water (0), and NH4C1 in fermentation media (I). These were diluted with water to given ammonium ion concen— trations of 16.16 ug ml_1 and 5.39 ug ml-l. After Nesslerization their absorption spectra were recorded between 380 and 500 nm. ABSORBANCE 29 0.8 0.4 r 380 1 400 16.16Ag/ml l i J 420 440 460 WAVELENGTH (nm) 480 30 2N NaOH were added to each tube and the color was allowed to develop for 15 min. The Nessler's reagent contained 4.0 g KI and 4.0 g HgI2 per liter of water. The absorbance was read at 400 nm in a grating spectrOphotometer. The accuracy of the ammonium ion concentration was about 15% but decreased at high concentrations due to dilution error. 6.2.5 Glucose. Glucose concentrations in fermentor samples were determined by the phenol-H2804 colorimetric method (Dubois, et aZ., 1955) as modified by Johnson et al., (1956). Interference by a— and B-keto acids, aldehyde and ketones and, in particular, pyruvic acid was shown by Mont- gomery (1961). Pyruvic acid was previously shown to be a product of E. coli growth (Landwall and Holme, 1977) however the low concentrations found in this investigation did not indicate it as an interfering agent. Standards were prepared from a stock solution containing 100 ug/ml d-glucose. This was diluted with water to give standards containing 5, 10, 20, 30, 40, and 50 ug/ml d—glu- cose. Samples were diluted with water to give a final glu— cose concentration between 5-50 ug/ml. Two ml aliquots of the standards, diluted samples, and water (blank) were placed in reaction tubes. One ml phenol reagent (50 g/l phenol in water) and 5.0 ml of concentrated H2804 were added and the tubes were vortexed. After the tubes cooled for 1 hr the ab- sorbance was read at 490 nm. Glucose concentrations were then determined from a standard curve. Accuracy was within :5% but decreased at high concentration of glucose due to 31 dilution error. Concentrations less than 5 ug/ml were undetectable. 6.2.6 Volatile Fatty Acids. Fatty acids were analyzed by gas chromotography according to a modification of the pro- cedure described by Supelco, Inc. (1975). Standards were initially prepared from a volatile acid standard mixture (14M). The standard mix contained 1 milliequivalent of the fatty acids C -C7 per 100 ml water. However, only acetic and l prOpionic acids were found in fermentor samples (Figure 4A). The long elution time of heptanoic acid (30 min) made the use of this standard mix inconvenient. Subsequently a stan— dard mixture of only acetic and propionic acids at 4.026 and 0.985 g/l respectively was prepared, shortening the analysis time to about 6 minutes per standard injected (Figure 4B). This solution was diluted with water to give standard solutions containing the acids at 1/2, 1/4, and 1/8 of the above concentrations. Two ml aliquots of each fermentor sample and standard were placed in screw cap tubes. These were acidified di— rectly by adding 0.1 m1 of 50% H2804 acids were extracted with 1 m1 ethyl ether. These were cen— (v/v in water) and the trifuged to break the emulsion and the water layer was frozen. The ether layer was then poured into screw-cap tubes con- taining enough anhydrous NaZSO4 ume of ether. This removed any dissolved water in the ether. to equal about half the vol- A gas chromatograph (18M) equipped with a thermal con- ductivity detector was used for the analysis. The injection 32 Figure 4. Volatile Fatty Acid Chromatograms. Peaks were identified as 1, solvent (ether); 2, acetic acid (3.04 min); and 3, propionic acid (4.54 min). (A) Fermentor sample. (B) Standard mixture of acetic acid (2.013 g 1'1) and propionic acid (0.493 g 1"). 33 .1 0 2 4 6 ELUTION TIME (min) 34 temperature was set at 160°C. The detector temperature was 170°C with a bridge current of 200 mA. The carrier gas was helium at a flow rate of 20 ml/min. The column was a 1/8" O.D. x 6' seamless stainless steel tube packed with 10% SP-lOOO on 100/120 Chromosorb W AW (15M). The column oven temperature was 135°C. Peak areas were determined using a CDS 111 integrator (19M). Chromatograms were recorded with a chart recorder (9M). Fourteen ul samples of the ether layer were injected into the column twice and the average peak areas were calculated. The concentrations of acids in the original samples were deter- mined from a plot of concentration versus peak area obtained from the standards. Peak areas were within :5% of average. 6.2.7 Non-Volatile Acids. Non-volatile acids were analyzed by gas chromatography using a procedure described by Supelco, Inc. (1975). Standards were prepared from a non- volatile acid mixture containing 1 milliequivalent in 100 ml of water of pyruvic, lactic, oxaloacetic, oxalic, methyl malonic, malonic, fumaric, and succinic acids (16m). The standard solution was diluted with water to give additional solutions containing 1/2, 1/4, and 1/8 the concentration of each acid. One ml aliquots of samples and standards were placed in screw cap tubes. Methyl ester derivatives were prepared by adding 0.4 ml of 50% H2804 (v/v in water) and 2 m1 lipOpure methanol, vortexing the mixture, and allowing them to sit overnight. One ml of water and 0.5 ml chloroform were added to extract the acid methyl esters. Fourteen ul of the chloroform 35 layer was injected into the column twice and the average area calculated. Peak areas were found to be within 5% of average. Initially the same SP-1000 column was used as in the volatile fatty acid analysis. This gave poor separation of the acids from the solvent peak. Better separation was found using a 6' x 1/8" O.D. stainless steel seamless column packed with 15% DEGS on 80/100 Chromosorb W AW (17M). A chromatogram of the standard solution using this column is shown in Figure 5A. Peaks were identified by injecting individual solutions of the acid methyl esters, prepared as above, and noting their elution times. Results are shown in Table l. The column temperature was set at 115°C; the injector and detector temperatures were both 155°C. The thermal conductivity bridge current was 200 m A and the helium carrier gas flow rate was 20 ml/min. The solutions of pyruvic and oxaloacetic acids both gave peaks at 2.41 and 4.48 minutes. Similar problems have been observed by others (Supelco technical service department, personal communication). The spontaneous decomposition of oxaloacetic acid to pyruvic acid has also been noted (Mont- gomery, 1961). These two acids were assumed to elute as given in Table 1. Separation of malonic and fumeric acids was poor. How— ever they were not found in any sample chromatogram (Figure 5B). It was noted that using a higher column temperature would reduce the trailing of the malonic/fumaric acid pair but gave poorer separation of pyruvic and lactic acids from the solvent peak. Figure 5. 36 Non-Volatile Acid Chromatograms. Peaks are identified in Table l. (A) Standard acid solution. (B) Fermentor sample. 37 | I l l I l l I 0 2 4 6 8101214 ELUTION T IME (min) 38 TABLE 1. Non—volatile Acid Peak Identification. Peak Numbers Correspond to Chromatogram Peaks in Figure 5. Peak Coumpounda Elution Time (Minutes) 1 solvent (chloroform) 0.48 2 pyruvic acid 2.41 3 lactic acid 2.91 4 oxaloacetic acid 4.48 5 oxalic acid 5.53 6 methyl malonic acid 6.39 7 malonic acid 9.01 8 fumaric acid 9.77 9 succinic acid 13.27 u unidentified -— aacids were analyzed as their methyl ester derivatives. 39 Acetic and propionic acids were also treated and tested as above but their response was small compared to the non— volatile acids even at high concentrations (4.0 g/l and 1.0 g/l respectively). Acetic acid eluted at the same time as oxaloacetic acid. Similarly propionic acid eluted with oxalic acid. The unidentified peak in the sample chromato- gram (Figure 5B) was attributed to propionic acid. 6.3 Experimental Microfiltration System and Procedure A process flow diagram of the microfiltration system is shown in Figure 6A, and a photograph of the system is shown in Figure 6B. A Microferm Fermentor system equipped with a 5 l fermentor (10M) was used in all experiments. The working volume was 32 but was measured at the end of each eXperiment. This system included temperature control, agitation, and aeration. The temperature was held at 37 : 0.5°C by cir- culating water, the impeller speed was 1000 rpm, and air volumetric rate was 2 l min.1 1 working volume-l (measured at 70°F and 14.7 psia). Aeration and agitation were assumed sufficient to prevent oxygen limitation. Sterile medium was stored in a 10 l carboy and agitated by a magnetic stirring bar. The carbon was vented to the atmos-_ phere through a filter packed with cotton to prevent contam- ination. The media was fed directly to the fermentor by a finger-type peristaltic pump (13M) through silicone rubber tubing. Air was humidified by sparging through distilled water and filtered through cotton packing before being sparged 40 Figure 6. Experimental Process Flow Diagram and Photo- graph. (A) Experimental process flow diagram. Symbols represent, A-antifoam syringe, B—pH measuring and reference elec— trodes, C-condenser, CR-pH chart recorder, cw-cold water, f—cotton- packed air filter, H—humidifier, I—impeller, M-magnetic stirrer, P-pump, PC—pH controller, V-valve. Photograph of the microfiltration system. 4]. z.¢:‘ > 0 I. b—-------— 0“ I i I | hzccp.» :a whicfidi (thmlcmu 42 ,. T, Jenna-no.- " i A”..-‘...'-' 1 5 'IOIloooooo.... _,.,.....’. a... .............. ‘ 000...... \E‘ In . ~g .. 43 into the fermentor. Gases leaving the fermentor passed through a water—cooled condenser and were vented to the atmos— phere. The pH was controlled by automatic titration with 5M NaOH. Autoclavable measuring and reference electrodes (7M) were mounted in the fermentor. A pH controller and strip chart recorder (11M) activated a rotary-type peristaltic pump (1M) when the pH became too low. pH was maintained at 7.0 i 0.05. The titrant reservoir was vented to the atmos— phere through a cotton—packed filter. Polypropylene glycol (molecular weight 2000) was used as antifoam, but was not adequate at cell concentrations greater than 1010 ml-l. Antifoam A concentrate (12M) was found more suitable. Either antifoam was added as needed through a syringe. The fermentation broth was circulated past a microfil— tration membrane (Acroflux Capsule, 6M). A photograph of the membrane capsule is shown in Figure 7. The membrane area was 1000 cm2 with a 0.2 pm pore size. Cell—free effluent was removed through a throttling valve. The circu— lation rate (filter crossflow rate) was typically 3.5 l min-1 but this was allowed to fluctuate in response to liquid level in the fermentor. A detailed analysis of the level control system is given in section 6.4; start—up procedures are given in Appendix IV. In an effort to minimize the effect of removing samples from the fermentor, 15 ml samples were taken at no less than 44 moocwflom cmsaoo m0 ammuusoo 3‘ . n .. .. 1. . I 1.10,}..kc. I . ’<.o."".‘.’ .' . .o.<-....‘l‘ ‘96:”. 1.442 . \ t . p ‘50; u.v1num, Figure 7. Photograph of the Acroflux Capsule 45 1 1/2 hour intervals during a microfiltration experiment. Experiments were continued until the membrane became fouled and filtration rates could no longer be maintained. Cleaning and storage of the filter was done according to the procedures of Tanny, Mirelman and Pistole (1980). The filter was removed from the system and washed by circulating a 95:5 ethanol-water mixture through the filter capsule for 20 min, then rinsing with distilled water. Severely fouled membrane capsules were washed with a 0.1 N NaOH solution for 1 hr and rinsed with distilled water. Filters were allowed to drain dry for storage and rinsed with sterile distilled water before subsequent use. 6.4 Level Control A closed loop level control system was developed for the microfiltration system as shown in Figure 8. Feed was supplied at a constant rate and filtrate flow rate was manip- ulated in response to process changes that would lead to fluctuations in fermentor level (i.e., pH titrant, filter fouling, etc.). Filtration rate has been shown to vary directly with filter differential pressure and shear rate tangent to the filter surface in similar microfiltration devices (Henry and Allred, 1972). Since both of these para- meters depend on the circulation rate past the membrane (crossflow pumping rate), the filtrate flow rate may be varied by changing the crossflow pump speed. The liquid level was monitored by a Teflon insulated probe and transmitter (5M) producing a 4-20 mA current in the 46 Emummfln mooq Houucoo H0>wq .m muomflm I I $05.2 I l _I I «E 00 J mi; 22?. l _ 02.5»R0m1» Beaummozo 65551.5: IOhZNEEwu zwhiu mhthmhwm $.30“. 47 control loop that is proportional to the fermentor level. The retentate inlet was directed away from the probe and the fermentor was sufficiently baffled to prevent vortexing and provide a smooth liquid surface. A strip chart recorder (9M) was used to give a continuous record of fermentor level. A current output controller (8M) in the control 100p generated a 4-20 mA response to the transmitter current. The response current controlled the crossflow pump speed (3M). Higher filtrate flow rates occurred than were required for the microfiltration system due to the large capacity of the microfilter and crossflow pump. The pump would run at very low speed resulting in a long residence time for cells in the filtration loop. Therefore, a throttling valve was added in the filtrate line to increase the pressure on the filtrate side of the filter membrane resulting in lower fil— tration rates, higher crossflow pump speed, and a short res— idence time in the filtration 100p. The level control loop could hold the volume to within : 5% of the total volume, however operation to i 1% was typical. 6.4.1 Control Loop Analysis. The control loop may be represented by the block diagram shown in Figure 9. Transfer functions are given as their Laplace transforms. An overall material balance around the system, assuming constant liquid density, gave for the process transfer function (Gp) G _ L(b) = l p \_ A b (23) Emummfla xooam Honucoo H0>0q .m wuomflm J. 48 G + u¢ , ¢ D/_\ Emu/V -wa 49 The controller is the proportional—integral—derivative type and its transfer function (GC) was given by G=1+TDo+—. (24) The motor response to control current is shown in Figure 10. Its transfer function (Gm) may be approximated by . 64.8 —9—r m ; if r > 9.2 mA R D ( > G = 7F: 25 m rpm . . ID 0 Z , 1f ID i 9.2 mA The transmitter response to level changes was linear over the active length of the probe and the transmitter gain (KT) is 0.764 mA cm—l. No measurement lag was observed. The filter transfer function Gf is a complex relation involv: ing the hydrodynamics of the filter capsule system. No attempts were made to determine this function. 50 600 l 500 ~ 200 - 100 - ID (mA) Figure 10. Pump Speed (R) vs. Controller Output Current (ID) RESULTS AND DISCUSSION The results of a control batch and the microfiltration experiments are given in graphical form and are discussed sequentially in Section 7.1. The mathematical model and computer simulation are presented and discussed in Section 7.2. 7.1 Experimental Results and Discussion The effect of microfiltration cultivation on the growth of Escherichia coli was investigated by studying the influ- ence of hydraulic residence time, TH (defined as Vf/Ff) and the concentration of glucose in the feed media, Sfo' A summary of these experiments is given in Table 2. TABLE 2. Summary of Microfiltration Experiments ' “'1 Experiment rH(hr) Sf0(g 2 ) l w (Batch) -- 2 4.15 3 2.94 8 4 1.58 16 The cell dry mass, viable cells, glucose, ammonium ion and acid concentrations were plotted as functions of time. Actual numerical data for each experiment is provided in Appendix III. 51 52 7.1.1 Control Batch Cultivation: Experiment 1. A batch culture was done for later comparison with microfiltra- tion experiments. Figure 11A shows the exponential growth phase through hour 11. The specific growth rate of cell dry weight and viable cells were 0.49 hr—1 and 0.50 hr-l respec- tively during this phase. The specific growth rates were cal- culated by a least squares fit of the slope from the plots of in Xf vs t and in Nf vs. t, that is from Equation 2 and Nf was assumed proportional to Xf. A gradual decrease in the specific growth rate followed the exponential growth phase, as is characteristic of glucose- limited batch growth of E. coli during the transition to stationary phase (Amarsingham and Davis, 1965). The simul- taneous depletion of glucose and ammonia is shown in Figure 11B. The medium was believed to have become limiting in glucose; the residual glucose shown was believed to be interference by cellular products present in the sample and reflects the non-specific nature of the analytical method. Accumulation of acetic and propionic acids are shown in Figure 11C. Succinic acid was also detected but chromat- ogram peaks were too small to integrate. Entry into stationary phase occurred just prior to hour 17. At this time, glucose was depleted, ammonium ion consump- tion ceased, and the pH began to increase. The fermentor 53 Figure 11. Results of Experiment 1. Control batch growth of E. coli HBlOl. (A) Cell dry weight concentration (xf) and viable cell concentration (Nf). (B) Residual glucose (Sf) and ammonium ion (Af) concentrations. (C) Volatile acid concentrations. Acetic acid (0) and propionic acid (O) con- centrations. 54 z. 30:0}..3 — 10' - — 10' 107 10.0 1 .0 33 .x r m. TIME (hr) 55 - 1.2 3 O _ I 8:. O 0 q Q 24 20 :4ao mas (hr) 56 1.4 I- 1.2 - _ O. 1 . 3 O 53 20.5520 h _ s 4. 0 O 0200 93 0.2 a 24 l 16 20 12 TIME (hr) 57 broth was subsequently titrated with concentrated HCl to control pH. The rise in pH was attributed to a shift in metabolism from glucose to acetate catabolism as shown by the decrease in acetic acid concentration to undetectable levels. While samples were not taken after 24 hours, the experiment was allowed to continue. The pH strip chart re- corder showed additional HCl was required after 24 hours, probably due to propionic acid catabolism. Finally the pH re- mained stable and no further HCl was used. A slight increase in cell dry weight during the acetic acid catabolism period was noted, however the error in the data may be larger than the indicated increase. The final cell concentration was 2.37 g dry weight per liter or 7.2 x 109 viable cells per ml. The accumulation of prOpionate in the medium appears as a peculiar result of glucose metabolism. Previous studies on intracellular enzyme activity has brought forth a plausible explanation for this phenomena. Amarasingham and Davis (1965) found no a—ketoglutarate dehydrogenase activity in E. coli cells grown anaerobically or in exponentially growing cells under aerobic condition. This enzyme provides the link between a-ketoglutarate and succinate in the tricarboxylic acid (TCA) cycle (Figure 12). They prOposed that, under anaerobic con- ditions, the TCA cycle Operates as a branched pathway. The oxidative branch leads to a-ketoglutarate and the reductive branch leads to succinate. It appears that this modification to the TCA cycle is operating in exponentially growing cells under aerobic conditions as well. 58 CO2 Pvnuvne ® \ \ OXALOACETATE ‘CETYL'°°‘ ' I I / / I MALATE CITRATE 4 I I I I 6 FUMARATE ISOCITRATE 4 l | I . 1 t SUCCINATE a: - KETOGLUTARATE / l " \ ar-KETOGLUTARATE / l -.J@§wnmcqys§.z" l GLUTAMATE PROPIONATE REDUCTIVE OXIDATIVE BRANCH BRANCH Figure 12. Proposed Modification of the TCA Cycle to a Branched Pathway for Exponentially Growing E. coli Under Aerobiosis (adapted from Amarsingham and Davis, 1965). The intact TCA cycle (--+) is modified to a branched pathway (——») in the absence of the a-ketoglutarate dehydrogenase system. 59 This branched mechanism is consistant with the results of the batch experiment. The lack of glutamate in the media indicates that its synthesis from o-ketoglutarate, formed by the oxidative branch, is required. The accumulation of prOpionic acid as an endproduct is prOposed to be the result of the reductive pathway to succinate and finally to prOpionic acid. A diagram of the prOposed mechanism is shown in Figure 12. The lack of data on the activities of key enzymes in these pathways during the course of the experiment makes this proposed mechanism speculative. The accumulation of acetic acid in the medium by E. coli, as in this experiment, was attributed to an oversupply of NADH2 (Doelle, Ewings, and Hollywood, 1982). The repression of cytlochrome a by glucose was also demonstrated (Hollywood and Doelle, 1976) which would prevent electron transport from NADH2 to oxygen. Thus it appears that acetate may act as a temporary electron reservoir when excess glucose is present with subsequent oxidation of acetic acid when glucose is depleted. The terminal oxidation of acetic acid requires an active TCA cycle. Amarsingham and David (1965) found that o-ketOglut- arate dehydrogenase was induced when sufficient acetate accumu- lated. The induction of a—ketoglutarate dehydrogenase led to a smooth transition in growth rate during the shift from glucose to acetic acid catabolism. 60 7.1.2 Microfiltration Cultivation. Culture behavior in the microfiltration system was studied by following the transient growth of E. coli using different feed rates (hydraulic residence times) and nutrient feed compositions. Experimental results are compared to the results obtained in the control batch culture. Cells were allowed to grow in the fermentor as an ordinary batch culture prior to the start of microfiltration. Since the microfiltration capsules were not sterile after their first use, a high density of cells in the fermentor was desirable to prevent overgrowth of contaminants that may have been present. The final samples from each run was streaked for isolation on plate count agar to determine the extent of contamination if any. Few contaminants were found compared to the large number of E. coli colonies. In all runs, microfiltration was started before the cell density reached 0.6 g 2-1 dry weight in an attempt to wash out acid end products that may be inhibitory to the organism and delay or eliminate the decrease in specific growth rate found in the batch experiment during the transition to acetate cata— bolism and stationary phase. The microfiltration system was first tested to assess filtration performance and tune the level control system. Examination of the filter capsule at the end of these tests indicated a heavy build-up of cells between the outer channel wall and housing wall (Figure 13). The outer channel wall was discovered to be slightly permeable to water (Dale 61 @ ms" HOLIE DRILLED PERMEATE SILICON RUBBER FILLING +———————- RETEN TATE “OWRMG SEAL///r 5 MNG : C I II C BYPASS » FLOW 4 . OUTER CHANNEL wALL ,4 HOUSING ” WALL j TURBULENCE-PROMOTING . SUPPORT SCREEN f U :I MEMBRANE r4 .2 (A (A / K r POTTING SEAL IF FEED Figure 13. Cut-Away View of the Modified Gelman Acroflux Capsule 62 Hauk, Gelman Sciences, personal communication) causing a net flow of the fermentation broth into this dead space. The problem was circumvented by drilling 3 x 1/16" I.D. holes through the housing wall from the tOp, equidistant around the capsule, and through the seal ring. The hole in the housing wall was later filled with silicon rubber cement. This created a bypass around the outer channel wall, eliminating the dead space, allowing the fermentation broth to sweep any deposited cells from the space. The by- pass flow did not decrease filtration rates through the mem- brane noticably. The filtrate was clear but took on a yellowish tint as the cell density in the fermentor increased. Viable cell T counts in the filtrate were taken during some of the pre- liminary experiments. Typically less than 100 cell ml-1 were found. This was considered negligible compared to the number of cells in the fermentor. All experiments were continued until the membrane be- came fouled and filtration rates could no longer be main- tained. Attempts to backflush the filter with filtrate were moderately successful, but soon afterwards, the filtration rate again became too low. 7.1.2.1 Lonngesidence Time Microfiltration: Experiment 2. A long residence time microfiltration experiment was per- formed in an effort to extend eXponential growth beyond that found in the batch experiment and prevent the accumulation of inhibitory acid. A new Acroflux capsule was used for the run. 63 Microfiltration was started 6 hours after inoculation. The feed rate was 592.5 ml hr—1 and the fermentor working vol— ume was 2460 ml giving a hydraulic residence time of 4.15 hours. Results from experiment 2 are shown in Figure 14. A characteristic drop in the fermentor cell concentra— tion followed the start—up (Figure 14A). This was attributed to adhesion of cells to the filter membrane surface. Tanny, Merelman, and Pistole (1980) observed the same effect, re- flected in low cell recoveries (44.5 to 76.1%), when using the filter to harvest small batches of cells. The layer of cells on the membrane is believed to attain a steady-state thickness as turbulence near the surface inhibits further deposition of cells. The subsequent accumulation of cells in the fermentor continued at a lower specific rate than in the batch portion of the experiment. The final cr0p of cells was 4.84 g £_1 10 viable cells ml-1 representing a dry weight and 1.4 x 10 two—fold increase over the batch culture. The consumption of glucose and ammOnium ion proceeded the same as in Experiment 1 indicating that the rate of utilization of these components by the culture was much faster than the rate at which they were fed (Figure 14B). Growth became glucose—limited as indicated by the depletion of glucose in the fermentor near the end of the experiment. The residual glucose shown is believed to be interference in the glucose analysis. Figure 14. 64 Results of Experiment 2. Microfiltration experiment with a hydraulic residence time of 4.15 hr. (A) Cell dry weight concentration (Xf) and viable cell concentration (Nf). (B) Residual glucose concentration (Sf) and ammonium ion concentration (Af). (C) Acetic and propionic acid concentrations. 100.0 10.0 x'(g/I) O 0.1 I— “You ~ r I— I T I p——— utcnonuunou I I l ————-———O 12 TIME (hr) 24 10 8 ° 1 (um/sum) N I 66 1:— BATcH—g J T J. I I ‘—_——_ MICROFILTRAT|0N______. 0.8 0.4 8 TIME (hr) 12 16 20 (I/B) ’V 67 $303026 >06 002323352 85 4. 3 O O O 0 A _ _ 2 I - w \ O (9. L n w m w R T m l 8 3 m v n. V a HO 5 e _ _ 1 _ \0 3 6. 2. 8 4. 1 1 cl 0 o O 25v 20:.(zhzwozoo 0.0< 02h": TIME (hr) 68 Acetic and prOpionic acid concentrations continued to increase throughout much of the experiment despite the diluting influence of adding fresh media (Figure 14C). The ultimate decrease in acetic acid was attributed to a shift in metabolism; from an aerobic fermentative metabolism, where acetic acid was the primary fermentation end product (Figure 15A) to a respiratory metabolism with the complete oxidation of glucose to CO2 and water (Figure 15B) with wash—out of acetic acid by microfiltration. This is in contrast to the decrease in the acetic acid concentration found in Experiment 1 which was due to a shift from glucose to acetic acid catabolism by the culture in the absence of glucose (Figure 15C). The propionic acid concentration continued to increase even after the acetic acid concentration began to decrease. It is not clear why this should occur based on the mechanism previously described for propionic acid production. The shift in metabolism from aerobic fermentation to res— piration in E. coli was previously demonstrated in turbidostat culture (Doelle, Hollywood, and Westwood, 1974). At glucose concentrations greater than 2 g 2_l, metabolism was charac— terized by a high specific acid production rate and no a— ketogluta—rate dehydrogenase activity. At low glucose con— centrations (< 1.5 g £_l) there was no acid production and q-ketogluta-rate dehydrogenase activity increased to high levels. It appears that the shift to respiration occurs as 6) © GLUCOSE ACETIC ACID GLUCOSE CELL C02+ H20 ACETIC ACID CELL C02+ H20 Figure 15. III g Metabolic Shifts in E. coli under Aerobic Conditions. (A) (B) (C) Excess glucose in the medium results in the accumulation of acids (par— ticularly acetic acid) as aerobic fermentation end products. When glucose becomes growth limiting, acids are not formed and glucose cata- bolism occurs by respiration. If glucose becomes depleted as in Experiment 1, acetic acid is terminally oxidized. 70 glucose becomes the limiting nutrient indicating a more prudent use of the energy source when it is in limited supply. In Experiment 2, glucose was depleted between hour 13 and 14. This coincided with the decrease in acetic acid con— centration in the fermentor. It was clear that the nutrient feed rate was not fast enough to satisfy the nutrient demand of the culture. 7.1.2.2 Moderate Residence Time Microfiltration: Experiment 3. A higher feed rate was used in an effort to provide more nutrients for the growing culture. A faster wash-out of acetic acid was expected upon entry into glucose— 1imited growth than was found in Experiment 2 due to the shorter hydraulic residence time. In Experiment 3, media was fed at 1008.1 ml hr_l. The fermentor working volume was 2960 ml giving a hydraulic res- idence time of 2.94 hours. An Acroflux capsule that had been used for several of the preliminary experiments was used for the experiment. The results are shown in Figure 16. Microfiltration was started at hour 7.3. The growth curves of cell dry weight and viable cell concentrations did not show the decrease en- countered with new capsules at start-up. It was concluded that repeated cleaning and deposits of old material on the membrane, and perhaps in the micrOporous matrix, in some way prevented new material from depositing on the membrane. ‘I Figure 16. 71 Results of Experiment 3. Microfiltration run with a hydraulic residence time of 2.94 hr. (A) Cell dry weight (Xf) and viable cell (Nf) concentrations. (B) Residual glucose concentration (Sf) and ammonium ion concentration (Af). (C) Volatile acid concentrations. Acetic acid (O) and prOpionic acid (O) concen- trations. 72 11 100.0 I I T I I I 10 .— urcu ——.Ip___——— Iicnonuunoa ———————*I o 0 o 0 10 10.0 - — 10 z < .4 a O V . — 10°; " R 2 o 1 — 10° 1 0.01 L L 1 1 107 o 4 e 12 16 20 24 TIME (hr) 73 I I T V W ‘— BATCH—.I—-—————MICROFILTRATION 1 a — 6 - q i 346'— -4 m o . 2 - .I o l l l l 0 8 12 16 20 24 TIME (hr) 0.8 O.‘ (IN) 'v ACID CONCENTRATION (g/l) 1.4 0.8 0.4 0.2 74 l I MICROFILTRATION——.' — TIME (hr) 75 Exponential growth continued until hour 10 when the specific growth rate began to decrease (Figure 16A). At hour 14, glucose became growth—limiting as seen by its de- pletion (Figure 16B). The ammonium ion concentration de— creased as growth proceeded but began to increase during the glucose-limited growth phase as would be expected for an excess nutrient. At the time microfiltration was started, the rate of glucose and ammonium ions entering the fermentor was less than the demand for those components by the culture. This resulted in time-concentration profiles similar to those in Experiments 1 and 2. Acetic acid accumulated in the fermentor during expo— nential growth phase but as growth became glucose—limited (between hour 12 and 14) the acid was washed out (Figure 16C). The decrease is much faster than in Experiment 2 due to the shorter hydraulic residence time in the fermentor. Propionic acid production continued throughout the experiment but accumulated to a lower level than in Experiment 2 perhaps due to the shorter residence time. The final cell dry mass and viable cell concentrations 10 cells ml.1 re— were 12.1 g 2-1 dry weight and 3.2 x 10 spectively, or about 5 times that attained in batch culture. Exponential growth continued at higher cell densities than in the previous two experiments. The nutrient feed rate however was soon unable to satisfy the ever increasing demand for glucose and growth became glucose limited. 76 7.1.2.3 Short Residence Time Microfiltration with Concentrated Feed: Experiment 4. In an effort to satisfy the demand for glucose which became growth limiting in the previous experiments, the feed was made with 16.0 g 2-1 glucose and a higher feed rate was used. The NH4Cl concen— tration was increased to 4.0 g £_1 to ensure that nitrogen remained in excess supply. The other medium components were at the same concentrations as in the previous experiments. The feed rate was 1773 ml hr—1 and the fermentor working volume was 2800 ml giving a hydraulic residence time of 1.58 hours. The results are shown in Figure 17. The same Acroflux capsule was used as in Experiment 2. Microfiltration was started just after hour 7. Adhesion of cells to the filter was evident in the Experiment but was not as pronounced as in Experiment 2. A decrease in the specific growth rate after start—up was also noted (Figure 17A). By hour 10.75 the filter became fouled and the high filtration rate could not be maintained. The capsule was re- moved and replaced with a previously used capsule to continue the experiment. The Viable cell concentration began to level off at about 1010 ml_1 after hour 12. It was believed that inhibition by accumulated acid prevented further in— creases in the viable cells. A combination of the short residence time and highly concentrated feed led to an increase in glucose and ammonium ion concentration at start—up (Figure 17B). The glucose de— mand by the culture eventually exceeded the feed supply Figure 17. 77 Results of Experiment 4. Microfiltration run with a hydraulic resi- dence time of 1.58 hrs. Glucose and NH4C1 in the feed was increased to 16.0 and 4.0 g 1—1 respectively. (A) Cell dry weight concentration (Xf) and viable cell concentration (Nf). (B) Residual glucose (Sf) and ammonium ion (Af) concentration. (C) Volatile acid concentrations. Acetic acid (0) and propionic acid (0). (D) Non—volatile acid concentrations. Pyruvic acid (0), lactic acid (I), ox- aloacetic acid (0), and succinic acid (0). 78 11 100.0 fi I I 1 -l 10 p— IATCH ———-On————-——- Incnonunnou ____——. >- .1 O 10 0 -— O _ 1I310 — o —1 z < 8‘ 31 0 — _ 10’ g " t x 3 0.1 — _ 100 t —1 0.01 i 1 k l 4 107 0 4 8 12 16 20 24 TIME (hr) ‘l 7‘9 4 I I — MICROFILTRAT ION a 12 TIME (hr) (vs) ’V ACID CONCENTRATION (g/I) 80 : T f T ' i—— BATCH —-—qI0——Nncaonunnlou _—-U 3 )— 2 _ 1 h. O 0\ M e 1 O 4 8 12 16 20 TlME(hr) 81 >O_O 0020m243>4.02 32v 3 2 4| 0 _ _ A r# I. N m T A R T L H 0 n l 1m I IN C T A B H _ _ _ o. s 4. 2. 1 0 0 0 0 53 295523200 99. 20 16 12 TIME(hr) 82 leading to its decline after hour ll. The ammonium ion con— centration also increased but remained at high levels throughout the experiment, indicating it was an adequate supply throughout. Glucose—limited growth did not occur and aerobic fer— mentation continued throughout the experiment. The utili— zation of the large amount of glucose in the feed by the culture led to the accumulation of larger amounts of acid than in the previous experiments. Figure 17C shows the accumulation of acetic and propionic acids during the ex— periment. Analysis for non-volatile acids, which were only weakly detected in previous experiments (concentrations less than about 0.1 g l_l) showed large amounts of oxaloacetic, lactic, pyruvic, and succinic acids in the medium. It was not clear why these acids should accumulate to high levels or even why they should be produced at all. 1 The final cell concentration was 12.43 g l— dry weight 10 viable cells ml"1 representing a 5 fold in— and 1.2 x 10 crease in cell dry weight and a 2 fold increase in Viable cells over the batch culture. 7.2 Mathematical Model and Computer Simulation 7.2.1 Mathematical Model. A mathematical model was developed to describe the transient growth of E. coli in the microfiltration system shown in Figure 1B. In all ex— periments the purge stream flow rate (Fp) was 0 and Ff = Fd. 83 The concentrations of fermentor components modeled were cell dry weight, Viable cells, the limiting nutrient, glu— cose, the excess nutrient, nitrogen (ammonium ion), and the end product acetic acid. The membrane was assumed to exhibit no selective permeation toward any solute in the liquid fraction of the fermentor broth thus the filtrate has the same composition as the liquid fraction in the fermentor. It was further as— sumed that the space occupied by the cells was negligible at the cell pOpulations found in these experiments. The highest packed volume of cells found in any sample was less than 8% of the total sample volume. The component material balances around the system gives for the cell dry weight (Xf), . (27) The viable cells per ml (Nf) were assumed proportional to the cell dry mass concentration, dN l f _ _ d—t— — 0.001 mc I'g (28) where Inc is the single cell mass and the 0.001 converts l to ml. The glucose concentration (Sf) is given by, (29) 84 the ammonium ion concentration (Af) is f _ l _ —dt — IA + i (Afo Af) (30) and the acetic acid concentration (Pf) is dP f _ l _ Unstructured kinetic models were used to describe the intrinsic rates of reaction. The simplest kinetic models were Chosen for which kinetic parameters were available or could be determined from the batch experiment. The rate of growth (rg) is given by the Monod equation; then from equations (2) and (4) Sf r=u ( 9 max §E¥Kg> X (32) f' Glucose is the sole carbon and energy source for the organism and the rate of glucose utilization (~rs) contains terms for both growth and maintainance. Then equation (9) may be written (33) The rate of ammonia utilization (-rA) as the nitrogen source is given by -r =0: r. (34) 85 The rate of acetic acid production (rp) is represented as a fraction of the substrate consumed by rP = y(—rs). (35) 7.2.2 Kinetic Parameter Estimation. Kinetic parameters were estimated from batch data or taken from the literature. Maximum Specific Growth Rate, Llmax--—Glucose is in large excess during the early exponential growth phase. equation (32) then reduces to r9 = “max Xf. (36) Substituting equation (36) into (27) gives 1 dxf _ d(ln Xf) _ i; dt_ _ —__dt__— _ umax’ (37) Comparison of equations (37) and (26) gives L1 = L1 (38) during early exponential growth phase. The early exponential growth phase specific growth rates were variable between experiments (Table 3). The specific growth rate from the batch experiment was taken as the maximum specific growth rate of 0.50 hr-l. 86 TABLE 3. Early Exponential Growth Phase Specific Growth Rates (hr Experiment Specific Cell Dry Specific Viable Weight Growth Rate Cell Growth Rate l 0.49 0.50 2 0.57 0.48 3 0.57 0.63 4 0.55 0.56 Substrate Saturation Constant, Ks-—No experiments were performed to determine this parameter. A value of 0.05 g l_1 was selected, however, any reasonable value does not alter the computer outcome significantly. In Single Cell Mass, c——Substitution of equation (27) in— to (28) gives 1 dxf 0.00]. By: mc. (39) f A plot of Xf vs. Nf from the batch data (Figure 18) gives dX . f = 3.74 x lO—lO g ml . Then from equation (39) de cells 1 _ -l 10 g ml _ -13 — mC — 0.001 ml x 3.74 x 10 cells 1 — 3.74 x lo g cell Glucose Consumption Coefficient, OLs—-Maintenance metabolism was neglected for this calculation. Substitution of equations (33) and (27) into (29) gives for batch growth dsf _ dxf dt ‘ ’“s dt ' (40) 87 T j Y 1.0 - O l l I I _ m A ISLOPE 33.70110 c.". I < ______ .1 a t-O.1 — x C 0 01 1 1 1 l I 7 1o 10' 10' N, (CELLS/m1) Figure 18. Cell Dry Weight (Xf) vs. Viable Cell (Nf) Concentrations from Experiment 1 88 This was simplified to AS f _ _Z—X—f_—GS. (41) The initial cell concentration from Experiment 1 was 1 and the final crop of cells was 2.34 g £_l. 1 0.0072 g R— The initial glucose concentration was 8.0 g 1_ and was 0 (neglecting interference in the glucose analysis) at the end of the experiment. Then _ _ 0-8.0 _ —1 OLs ‘ 2.34-0.0077 ‘ 3°43 9 9 Maintenance Metabolism Parameter, B——Comparison of Equations (9) and (33) shows that % = as and a— _ y — as a - B (42) Marr, Nilson and Clark (1963) found the value of a to be 0.025 hr_l for E. coli. Then using this value in equation (42) gives 1 l x 0.025 hr‘ = 0.086 g g’1 ’1. B = as a = 3.43 g g" hr Ammonium Ion Consumption Coefficient aA—-This parameter was calculated in a manner analogous to us. That is AA f _ = a (43) AXF A The initial ammonium ion concentration was 0.0876 g 2—1 and the final was 0.276 g 1—1. Then from equation 43 89 _ _ 0.276-0.876 _ -1 0‘A ‘ 2.34-0.0072 ‘ 0'26 9 g Acetic Acid Yield Coefficient,y-—Substituting equations (29) and (31) into (35) gives for batch growth de dS dt_ = Y(- -——). (44) This was simplified to f - -—— = Y (45) ASf The initial acetic acid concentration was 0 g 2—1 and the 1 final concentration was taken as 1.47 g 2— before terminal oxidation began. Then y = —4-—6— = 0.183 g g‘l. A summary of the kinetic parameters are given in Table 4. 90 TABLE 4. Kinetic Parameters for the Mathematical Model Symbol Value Probable Error (%) 0 o 50 hr'1 26 max ' Ks 0.05 g 2'1 9o mC 3.74x1o‘l3 g(ce11)'l 30 as 3.43 g g_1 20 8 0.086 g g'lhr’l 50 -1 0A 0.26 g g 20 Y 0.183 g g-1 20 7.2.3 Computer Simulation. The model was used to sim— ulate the transient growth of E. coli in the microfiltration system using the parameters estimated from the batch data. Step—wise integration by digital computer using the Second Order Runge-Kutta Method with a step size of 0.01 and 0.001 hours. The error associated with the smaller step size is 100x less than with the larger step size. The same result was obtained with both step sizes. The computer program and output are given in Appendix V. The kinetic equations do not predict a lag phase; there— fore the initial conditions were taken from a point in the experimental data early in exponential growth phase. Physio— logical changes within the cell, such as the transition from aerobic fermentation to respiration, are not reflected in the unstructured kinetic model. 91 The maintenance term in Equation (33) can cause Equation (29) to predict negative glucose concentrations. To remedy this unphysical situation, all of the reaction rate terms were set to 0 if the glucose concentration became less than or equal to 0. The simulated results from each experiment are discussed sequentially. 7.2.3.1 Batch Growth Simulation. The batch growth of cells was closely predicted (Figure 19A). The model did not predict the decrease in specific growth rate after hour 11 since the kinetics did not contain information regarding the shifts in metabolism encountered with this organism. The consumption of medium components was directly coupled to the rate of growth in the model. The growth rate of cells was predicted faster than was found experimentally, therefore the rate of glucose and ammonium ion consumption was also predicted faster than was found experimentally (Figure 19B). The accumulation of acetic acid was similarly predicted faster than was seen in the experiment (Figure 19C). 7.2.3.2 Experiment 2 Simulation. The simulation of cell growth in Experiment 2 was only moderately successful. The specific growth of cell dry weight during the batch portion of the experiment was faster than in the simulation. The deposition of cells at the start of microfiltration was not included in the model but was severe in this experiment. Figure 19. 92 Computer Simulation of Experiment 1. Simulated results (-——) are compared with experimental results (———). (A) Cell dry weight (Xf) and viable cell (Nf) concentrations. (B) Glucose (Sf) and ammonium ion (Nf) concentrations. (C) Acetic acid concentration. 93 A 10' 2‘ 30:63.: —- 10' 107 10.0 1.0 53 .x TIME (hr) 94 >13: 2 8 4. 1. O 0 0 I— 1 — . B_ . 9 . . . _ . _ . . _ . . . . L . _ . v \\\ \\ J \ n. \H x \\ \ q. . \O l 0 0 L . O 0 - _ u _ 8 q 6 4 2 :3. .m 16 20 24 12 TIME (hr) 95 _ _ 1.4 '- 1.2 — 1.0 - 8 0 53 20.555. 6. 0 0200 0.0 4 0. < 0.2 TIME (hr) 96 The subsequent growth of cells during microfiltration was at a lower specific rate than in the simulation (Figure 20A). The predicted consumption of glucose and ammonium ion was faster than in the experiment (Figure 20B). The simulation shows an increase in the ammonium ion concentration as growth becomes glucose limited. This behavior was expected for an excess reactant, however the scatter in the ammonium ion concentration data near the end of the experiment failed to show the increase clearly. The accumulation of acetic acid was predicted accurately however the concentration increased to higher levels than in the simulation (Figure 20C). This is probably due to un— certainty in the value of Y in equation (35). The model predicts a constant production rate of acetic acid during glucose-limited growth. The data shows, however a decrease in the acetic acid concentration which was attributed to a shift in metabolism from aerobic fermentation to respiration (rp + 0) and wash—out of the accumulated acetic acid. 7.2.3.3 Experiment 3 Simulation. The growth of cells in Experiment 3 was faster than in the simulation, however close agreement between the model and the experimental data was found (Figure 21A). The consumption of glucose and ammonium ion was pre— dicted accurately (Figure 21B). The increase in the ammonium ion concentration during glucose—limited growth, which was expected for an excess reactant, was found in both the sim— ulation and data. Figure 20. 97 Computer Simulation of Experiment 2. Simulated results (———) are compared with experimental results (———). (A) Cell dry weight (Xf) and viable cell (Nf) concentrations. (B) Glucose (Sf) and ammonium ion (Af) concentration. (C) Acetic acid concentration. 100.0 10.0 XI(3/I) 0.1 98 11 T I 1 10 ______———-U .— uTcN —U)l——_ IICIOFILTRATION — 10” Z 8 -—4 10. : a \ 2 — 10' 1 I l 1 107 0 4 1s 20 24 12 TIME (hr) Sf (9/1) 99 l l I :— BATCH—$‘————— MICROF ILTRATION — 0.8 .3’ 4 04 a 2 0 20 TIME (hr) 100 J IN. 1 w T. A R m F 0 R IV. D N r n H m A .@ It? _ b c _ 0 s. 2 a 4 1 1 1. 0 0. 0 58 295523200 90¢ 0.59. 16 20 12 TIME (hr) Figure 21. 101 Computer Simulation of Experiment 3. Simulated results (—--) are compared with experimental results (———). (A) Cell dry weight (Xf) and viable cell (Nf) concentrations. (B) Glucose (Sf) and ammonium ion (Af) concentrations. (C) Acetic acid concentration. 102 \ 100.0 I 6@ ' I j aATCN ——-OIn———— MICROFILTRATION ——————II I 10.0 E < a v 1. —I ~ x 0.1 _ 0.01 l l l l I l 4 8 12 16 20 24 TIME (hr) 10 11 ’N (|u.I/S||:O) 103 sfig/n r I 1 1 t———-—mcnonunnuou TIME (hr) U/OI'V ACID CONCENTRATION (g/I) .a o .0 m 9 0'» .0 a. 0.2 104 I I 1 1 1 MICROFILTRATION —___—_—4.. TIME (hr) 105 The accumulation of acetic acid shown in Figure 21C was faster than was found in the simulation. The shift from aerobic fermentation to respiration was not included in the kinetics and hence wash—out of acetic acid was not seen in the simulation. 7.2.3.4 Experiment 4 Simulation. The growth of cells in Experiment 4 was predicted accurately through hour 12 (Figure 22A). Thereafter the growth of viable cells stopped even with an excess of glucose in the medium. The large amount of acids that accumulated during this experiment were believed inhibitory to the organism. Acid inhibition was not included in the model hence exponential growth was predicted beyond hour 12. The increase in glucose and ammonium ion concentration at the start of microfiltration was reflected in the simu- lation (Figure 22B). Glucose-limited growth was predicted to occur at about hour 13 but glucose was in excess through- out the experiment. This was the result of the inability of the model to predict the decrease in specific growth rate after hour 12. The ammonium ion concentration behavior was also not predicted after hour 12. The low specific growth rate found experimentally imposed a lower demand for ammonium ion as a nutrient after hour 12 whereas the simulation pre— dicted a high demand for this nutrient. The accumulation of acetic acid was also predicted up to hour 12 (Figure 22C). The high growth rate predicted Figure 22. 106 Computer Simulation of Experiment 4. Simulated results (——-) are compared with experimental results (———). (A) Cell dry weight (Xf) and viable cell (Nf) concentrations. (B) Glucose (Sf) and ammonium ion (Af) concentrations. (C) Acetic acid concentration. 107 11 100.0 I I I I I ,0 BATCH MICROFILTNATION , ’ -( 10.0 __ 10‘° C 31. — 10' X“ 0.1 —~ 10' 0.01 J l I I 1 107 o 4 a 12 16 20 24 TIME(hr) (Iw/Sllea) IN 108 l T MICMFILTRATION flAff‘I-l nan-y. r _ _ 10'- 8 5 :8. .m 20 16 12 8 TIME (hr) ACID CONCENTRATION (g/I) 109 Infra l MICROFILTMTION —. 1 l TIME (hr) 110 after hour 12 resulted in the acetic acid production rate to be predicted faster than was found in the experiment. The success of a mathematical model for a system under- going chemical reaction lies in the understanding of the reaction kinetics and biological systems are no exception. Until recently the only kinetic models in general use were of the unstructured type that only truely apply during balanced growth when all nutrients are in excess. They fail to account for variations in metabolism as was found in this investigation. CONCLUSION 8.1 Equipment Considerations Continuous fermentation with microfiltration appears as a promising method for increasing cell population den— sities and hence metabolite productivity per unit volume of fermentor. The application of microfiltration using microporous membranes has become a popular technique in affecting cell—liquid separations. Membrane fouling appeared as a major obstacle in run— ning the microfiltration system for long periods of time. Removal of the Acroflux capsule from the system for cleaning proved to be a satisfactory method for rejuvenating the membrane. Backflushing the membrane with filtrate in situ was only moderately successful. This problem should be more thoroughly studied before scale—up to an industrial process is attempted. A closed loop liquid level control system was developed to aid in maintaining constant fermentor working volume. The level transmitter and controller could be used directly on any size process fermentor; only the pumping horsepower need be considered for process scale-up. Similarly the control system could be applied, with modification, to other fermentor designs such as dialysis fermentation. 111 112 8.2 Metabolic Considerations The aerobic metabolism of E. coli was found to undergo several changes during the experiments. General trends in the experimental data along with studies on the metabolism of E. coli by previous investigators allowed some hypotheses to be formulated regarding these apparent changes. The key factor affecting these changes was the induction and repression of the d-ketoglutarate dehydrogenase system under the influence of glucose concentration. When glucose was in excess, this enzyme system was repressed and aerobic fermentation proceeded with the simultaneous production of acids, primarily acetic acid; the other TCA cycle enzymes then took on a biosynthetic role. Depletion of glucose in batch culture led to the induction of the enzyme system and terminal oxidation of accumulated acetic acid. In microfil— tration experiments, where glucose became growth limiting, the induction of the enzyme system led to a fully respiratory metabolism. Acetic acid that accumulated during glucose- excess growth was washed out by the microfiltration system. The formation of propionic acid was found as the result of the TCA cycle becoming a branched pathway during glucose- repression of d-ketoglutarate dehydrogenase. PrOpionic acid was the end product of the reductive branch to succinate. The regulation of propionic acid formation was not under- stood in this investigation however. 113 While it is not clear why these metabolic changes should occur, their results were apparent in this study. Further study into these mechanisms is warranted as they are strongly coupled to the growth process of the organism. 8.3 Cell Productivity Considerations The microfiltration system developed in this inves— tigation produced populations of Escherichia cOZi up to 5 times that obtained in the control batch culture. All ex— periments were terminated before a stationary or maintenance state of the culture was achieved due to membrane fouling. With better membrane performance, higher populations could be obtained. High concentrations of glucose in the feed was used in an effort to extend exponential growth phase and produce dense cell populations. This led to high residual glucose concentrations in the fermentor with the production of large amounts of organic acid end products which became inhibitory to further cell growth. The accumulation of these acids could be avoided by maintaining the culture in a state of glucose- limited growth so that a respiratory metabolism prevails. Growth could then continue in a linear fashion to produce high population densities. 8.4 Modeling Considerations The unstructured kinetic model predicted the general features of the time—concentration profiles found during 114 glucose excess growth, however, the kinetics did not include parameters for the shifts in metabolism encountered at low glucose concentrations. A structured model would be more appropriate in this case and the elucidation of the metabolic mechanisms of the organism is a first step in formulating these models. RECOMMENDATIONS 1. Measurement of the activity of certain enzymes involved in metabolism would be helpful in the further understanding of the data. 2. Assessment of the nutritional requirements of the organism and inhibitory effects of end products should be done before any long range study is attempted. 3. Background interference in the glucose and ammonium ion analysis was troublesome and these nonspecific pro- cedures should be dispensed with in favor of techniques such as high performance liquid chromatography (HPLC). 4. Continuous stirred tank reactor (CSTR) cultivation and the transient phenomena that occurs there can be a val- uable tool in determining kinetic parameters and studying microbial physiology in general. 5. Further experiments that maintain a low residual glucose concentration in the fermentor and hence a fully respiratory culture, could eliminate product inhibition and produce more dense cultures than obtained in this study. 115 APPENDICES APPENDIX I Standard Data Standard curves from the analytical procedures are given in graphical form. A least squares fit to a straight line was used to calculate concentrations from the standard data. Lines are drawn on the figures to illustrate cal- culation procedures (Appendix II). 116 ABSORBANCE (490nm) 0.8 0.6 0.4 (12‘ 117 l I I I 10 20 30 4O GLUCOSE CONCENTRATION (’19 /m I) Figure Al. Glucose Standard Curves Experiment 1 (0), Experiment 2 Experiment 3 (0), Experiment 4 (A) , (U) 50 Figure A2. 118 Ammonium Ion Standard Curves. Experiment 1 (0), Experiment 3 (0), Ex— periments 2 and 4 (I). The curves for Experiments 1 and 3 were obtained at 490 nm. The curve for Experiments 2 and 4 was obtained at 400 nm by the procedure in Section 6.2.4. Note the increased sensativity at 400 nm. ABSORBANCE 0.6 0.5 0.4 0.3 0.2 0.1 I I 10 2O AMMONIUM ION CONCENTRATION (pg/m1) AREA COUNTS x10'5 —A M 4 A I _I I Ismpeuzzsnos A"? ' ___ _ J 2 1 A 3 q l l 1 I 1 0 02 04 QB 03 1B 120 ACETIC ACID CONCENTRATION (g/I) Figure A3. Standard Curves for Acetic Acid Experiment 1 (0), Experiment 2 (A), Experiment 3 (0), Experiment 4 (D). The signal attenuation to the integ— rator was 4 in Experiment 4 and 8 in the others resulting in the increased sensativity. AREA COUNTSx10'5 b N 121 I I I I OA 06 02 to PROPIONIC ACID CONCEN 08 TRATION (g/I) Figure A4. Standard Curves for Propionic Acid Symbols and attenuation settings are the same as in Figure A3. 122 AREA COUNTSx1O‘5 .AOV pflom Oflcfloosm paw .ADV pflom Ufluomq .Alv pfiom oapmomoamxo .AOV papa OH>5Hmm 0 ucwEHHmmxm ROM m0>uso pumpcmum pflom mHflumHo>Icoz .mm musmflm ACE zo_h<¢h2wozoo o_o< Qo no so No o _ — H _ _ HIIIIIIII N n spIXSanOOVBHV APPENDIX II Sample Calculations Examples of calculations from raw data to yield the concentration of various sample components are shown for each analytical technique. Sample 10 of Experiment 4 is used to illustrate these calculations. OPTICAL DENSITY. One ml of sample was diluted to 100 fold with distilled water. The optical density of the diluted sample was found to be 0.177. The optical density of the original sample was then 0.177 x 100 = 17.7 O.D. CELL DRY WEIGHT. The optical density was multiplied by the slope of the standard curve in Figure 2. This gives 17.7 O.D. X 0.3289 = 5.821 g/l. g 2 O.D. VIABLE CELL COUNTS. The sample was diluted 108 fold in saline dilution blanks. The plate counts for the three plates were 132, 132, and 134 for an average of 132.7 colonies (unusually close agreement). The viable cells per m1 of original sample was then 132 x 108 = 1.3 x 1010 cells ml"1 GLUCOSE. One ml of sample was diluted 500 fold. After the diluted sample was treated as described in section 6.2, 123 124 the absorbence was measured to be 0.210. From the standard curve in Figure Al, this corresponded to 14.51 ug/ml. The glucose concentration in the original sample was m1 9 14.51 ug/ml x 500 x 10'3 Eng = 7.255 g/I = 7.26 g/I. AMMONIUM ION. One ml of sample was diluted 100 fold and Nesslorized as in section 6.2. The absorbance was found to be 0.395. From Figure A2 this corresponds to 8.76 Eflfi%flii The ammonium ion concentration is then w. -am___19= 2 8.76 ml x 100 x 10 2 pg 0.876 g/R 0.88 g/I. VOLATILE FATTY ACIDS. The integrator reported two peaks. The sample was injected twice and the average area was cal— culated. For acetic acid the areas were 953,128 and 942,674 giving an average of 947,901. The propionic acid areas were 194,608 and 208,472 giving an average of 201,540. The standard curve for acetic acid did not extend beyond an area of 250,000. Linearity was assumed in calculating the acetic acid concentration for this sample. The slope of the AREA COUNT Q 9 standard curve was 4.226 x 10 giving the acetic acid concentration in the sample as 947901 AREA COUNTS 1 4.226 x 105 §5§§—§99§3—£ = 2.24 g/I. The propionic acid concentration was read directly from Figure A4 giving 0.276 g/fi. 125 NON-VOLATILE ACIDS. These were handled similarly to the volatile fatty acids. The raw data are shown in table Al for sample 10. The concentrations were read from Figure A5. TABLE Al. Computation of Non-Volatile Acid Concentration Injection Injection Average Acid 1 2 Area Concentration (g/l) Pyruvic 17,880 16,624 17,252 0.120 Oxaloacetic 58,352 62,520 60,436 0.194 Lactic 34,428 27,912 31,170 0.415 Succinic 25,784 N.R. 25,784 0.0415 N.R.—no integrator report APPENDIX II I Numerical Data Numerical data is given in tabular form for each ex— perimental run. Missing data is indicated by a hyphen, ———, and was usually the result of lack of enough sample for analysis or some other experimental error. In Exper— iments 2 through 4 the starting times of microfiltration is noted. 126 127 oo~.o o.o FmN.o PN.Q mopxm.n ooo.~ wo.o «N Fm NmN.o o.o mwN.o NN.o moexm.o mmm.~ MN.~ NN ow co~.o nmo.o NwN.o mN.o mo—xm.n mwm.w om.m om mF mwr.o owo.w w-.o mm.o mo—xq.m mum.w mo.n me mp «MF.D wmc.e w-.o m—.o morxm.w mwM.N ON.~ NP up mnno.o noc.v mnm.o ww.o mo—xq.m Nan.~ Nr.m or or Fwwo.o smr.e wmm.o nm.~ oowa.c mmm.w cm.m mp mp muco.o memm.o mom.o mn.m moexc.c mow.e mm.a «F «— Nmmo.o onwm.o mon.o mn.¢ mo~xm.~ mmo.P o~.m m— mp mono.o mcmq.o In mm.w moexn.m «orm.o cmq.~ NF NF mneo.o wNon.o mNn.o mm.m moexc.r Owam.o wee.w PP PP 0.0 mFmF.o mpm.o mN.m mopxo.m Nqnm.o ero.— or or o.o «mow.o II an.m mowxm.q ncmw.o Nmm.o m m o.o nomo.o mnw.o n~.n mowxr.m mmFF.o qwm.o w m o.o nmqo.o wwm.o ww.n mowxm.e onno.o qmm.o m m o.o mmmo.o I: Fm.n moFxN.F nm¢0.o nme.o m m 9.5 o.o www.o NN.m norxo.w ooNo.o Fmo.o n m o.o 0.9 mam.o mr.m worxw.¢ quo.o omo.o a c o.o o.o I: mn.n no—xw.m m—wo.o mmo.o m m o.o o.o Nmm.o NN.m wopxm.e whoo.o NNo.o N N o.o o.o wmm.o mm.n morxa.m mmoo.o meo.o F F o.o o.o In Nq.m woaxm.m Nwoo.o NNo.o o o :5 9 l1 s $-15 :8 s 32 331 $4 9 :la 9 2:5 232.. 08A; 25 32.3 oflcofldoum ceamo< quz mmoonau manmfl> xpo Hamu med» .A8 u Iev cam Loamm .F acmeflumaxm hoe mama Hmoflumesz .N< w4m

xuo Hamu meh _.u; ma.a u :9 can; cam coaumaufifltoaoflz .N reusaamaxm you mama Hmoaame:z .m< msm xno Hamu oEH» .p; am.~ u IF can: as: conumauaatoaoaz .n ucmsaamaxu 881 arms Haoaamsaz .q< mam

:uxm ofiuowg oflcfluunm 00co5000a oqguo< +022 omooafiu ofinmw> 50o fiucu 0555 .0: mm.e u :5 :05: com coaguuafiwuouowz .q glueduoqu you wane #005005327 .mw u4m

mq .om musmflm 04) our 1] thh.3m2(¢h 0 . n 1‘ 54405200 _10 w 000 a. o 10...... J... . II 0. Gun. '90 O p r. . 0.0 0 o . be 0 o < 0. (£00. 5 o 2 0 A WHHHHHHHLH . 005 5. 0 mar 00 0 65 m 0 "m . 0.02 on. 0 o . 0 cm. 0 0 mm 0! C000 0 o mm :3 an. Ql)>lo 5 o 002... o - o 0 00—1-— N . é Cwocoowc hCCtO 133 (d) FILTRATE THROTTLING VALVE - CLOSED 2. Turn on power to crossflow pump and controller. This activates the entire control loop. Pump will not be turning. 3. Using the + button on the controller, hold down until pump turns at 250 rpm. 4. After the filtration loop is filled with liquid, adjust setpoint until all error lamps are off. (This will occur at or near a setpoint valve equal to the percent reading on the transmitter ammeter.) 5. Turn on the feed pump and set the desired flow— rate. Open the filtrate throttling valve so that the fil— trate flow rate appears to equal the feed rate. 6. Observe the level in the fermentor making adjust- ments on the throttling valve as needed*. When the system looks like it is operating satisfactorally, switch to auto— maticcnmthe controller. 7. If the system settles in above or below the original level adjust the setpoint appropriately to increase or decrease the level*. 8. During the coarse of a run the pump speed will generally increase as the membrane becomes fouled. When full speed is reached, Open the throttling valve slightly. This will allow the pump to slow down. * Any changes made in the system should be small with a sufficiently long waiting period to allow these changes to settle out. Appendix V m uter Pro ram and Simulated Results Computer Program and Output )X - I 5 I F. PI 2 I) E .L M X/ I 8G T II E F . C AX N I 5 E . . D XI I TTT 8) 5 DUB .L E n t v I / R 2’) FG A . . 51‘ . M 222 I M ) /// . N m G . )1I) 8L 1 . 1 NNN .M T A A 1 ) FFF T I / A G A F N SAP N FS R . H . D. _ _ . E NL T A P F 3 S R 000 M IL L A L N . T0 . FFF I E I H A . 9 :6 N SAP R XC F P . F F T ) A ((( E ) ( 0 L S D. n 0 D. R n t s D. 1 I ) R A A . 3 TT R . .HHH X I; I C . SH SF . A . N 0000 E .F H I 5 RP NA F9 0 A S m+++ IX) T M A EL 0. R PF NGP R TTTR NNN T IL W H TA IF 5 U . l 0 0 . DDD LS 1SAD. X T ./ 0 T P E. TS G 0 F2 I)T SS tntNN /RRR E NXG R R LN MA\.II . N HNA . T55 RN 111/1160 T+++ NO E71 G A AF ATSDF I E .1 ATT .0 it; D); 2 M .I T .N RE.NX D FHF1 R.. GI .FFF T tFFF 0 II. H S A . A300 . A LTSE TTH RT SAPNATCSAD. To REX C TF P.1CS E A).. LLD .A ...FUDM.._ T EMS T EN SG T H HTF3 I . U FU OOOXOu/OOO 0 PI. A I B. CKQL) NN. FTO A X0 1FFF E FFF GO XTI B . w I .AO E YI.O O F T .E /SAPN 22SAP G EI) I 1 X TX:I7 L RRF1 RRA . F T(((FR//((( P I R .. A. EA T: B EPXF C0. 5 SR 000-50))... 0) .XH X 6 L S NM6ID AI VN.. I O T (0 .HHH(TNNHHH1 TP X2( 4 1 E K IU.NN T T E.T1 MOF ))CTTCDDDCCGGDDD+ SO 91: 3 F D S KMOIE 12345 N 0) .S GSICDM+++IERR+++T TT ... 1:) W.X1 O )1 . TOOOOO OI TEO7 TO) OOTI./$APTR++SAPN =5 IIX IIII SA. DOGUO N99999 0R N 51 R .OH99EDGGRRRERGGRRRU TT 103 0 OR KME 1A14A1 I ..... 5P 1 IT.F AE1D..NERR(((NORR(((0 .O II1)IIIH I U4 2E.(E. R0000. 0N OOOORN:(O TN./..IR+++++IC K FT4 ((( A "7 =R.TR: D. T=O . . .RU(T=FS .5. .IKPFFFFFK +++++ :TIG TTT)TTTI R . P (A (1 EEEEEONT..OOO DEATI H11EE XNSAP FFFFFTD .0 AAALAAA. 6 L3 0 DM 0+ TTTTT.INO :00 KTMN DD=TTL =====L XNSAPN+ TTPMMM/MMM2 0 A=OT AR AI IIIIIORU==OOO (IRUD (AUIIL NNNNNL =====UT ( ORRRGRRR.D ROEC=S E0 E: RRRRR=P0HTFFF FROON FEARRA FFFFFA FFFFFO: FOT000(0004N PCRMIT RF RI WWWWWTNKDDASP IWFKE IRTWWC XNSAPC XNSAPKT IGSFFFUFFFWE O O O O O 0123 456 1 2 4 5 6 7000 000 999 999 C C C C C C C C 1234567890123456789012345678901234567890123456789012345678901234567 11111.11111222222222233333333334444444444555555555566666666 134 400.04.... 135 Table A6 (cont'd). R6 R5 RA p) th15 ALPHAA GAMMA fl (’1 AMMA'RS COOOOC'I ....n0> Zr‘ mAuM0ommqmwhmn‘ n 2 mnnnmnnmn va>0mmv> U-‘lll II ll "—1" II II II 136 22922.56 (cont'd) EXPERIMENT 1 TIME xr NF SF AF PF (HR) (G/L) (CELLS/ML) (G/L) (G/L) (G/L) BATCH GROWTH 0.0 012 .26E+08 8.000 890 0.000 .5 015 .356+08 7.988 889 .002 1.0 019 .466+08 7.973 888 .006 1.5 024 .6OE+08 7.964 887 .008 2.0 031 .7BE+08 7.930 885 013 2.5 040 .1oe+09 7.898 883 019 3.0 051 .136+09 7.857 880 026 3.5 065 .17E+09 7.806 876 036 4.0 084 .226+o9 7.739 871 .048 4.5 108 .28E+09 7.654 865 .063 5.0 138 .36E+09 7.546 857 083 5.5 177 .47E+09 7.404 847 109 6.0 227 .60E+09 7.225 834 142 6.5 290 .77E+09 6.995 817 .184 7.0 372 .99E+o9 6.700 796 238 7.5 .477 .13E+1O 6.322 769 307 8.0 .611 .1GE+1O 5.839 734 396 8.5 .783 .21E+1O 5.219 689 509 9.0 1.003 .27E+1O 4.427 632 654 9.5 1.284 .34E+1o 3.414 559 .839 10.0 1.641 .44E+1o 2.126 466 1.075 10.5 2.086 .565+1O 521 351 1.369 11.0 2.227 .60E+1O -.000 314 1.464 11.5 2 227 .60E+1O - 000 314 1.464 12.0 2.227 .60E+1O —.000 314 1.464 12.5 2.227 .GOE+1O - 000 314 1.464 13.0 2.227 .60E+1O - 000 314 1 464 13.6 2.227 .60E+1O - 000 314 1.464 14.0 2.227 .60E+1O —.000 314 1.464 14.5 2.227 .60E+10 —.000 314 1 464 15.0 2.227 .60E+1O -.000 314 1 464 15.5 2 227 .GOE+1O - 000 314 1 464 16.0 2.227 .60E+1O - 000 314 1 464 16.5 2 227 .GOE+1O — 000 314 1 464 17.0 2.227 .60E+1O -.000 314 1 464 17.5 2.227 .6oe+1o - 000 314 1 464 18.0 2 227 .GOE+1O -.000 314 1 464 18.5 2 227 .60E+1O -.000 314 1.464 19.0 2.227 .GOE+1O -.000 314 1.464 19.5 2.227 .GOE+1O ..000 314 1.464 20.0 2 227 .606+1o - 000 314 1.464 20.5 2 227 .606+1o ~.ooo 314 1.464 21.0 2.227 .60E+1O - 000 314 1 464 137 Table A6 (cont'd). EXPERIMENT 2 TIME xE NF SF AF PF (HR) (G/L) (CELLS/ML) (G/L) (G/L) (G/L) BATCH GROWTH 0.0 .034 .54E+08 8.000 620 0.000 .5 044 .808+08 7.965 617 .006 1.0 .056 .11E+09 7.921 614 015 1.5 .072 .166+09 7.864 610 025 2.0 .092 .21E+09 7.790 605 038 2.5 118 .28E+09 7.697 598 056 3.0 .152 .37E+09 7.576 589 078 3.5 .195 .486+09 7.422 578 106 4.0 249 .63E+09 7.224 564 142 4.5 .320 .82E+09 6.971 546 188 ———4 5 -START MICROFILTRATION - RESIDENCE TxME= 4.15 HR- 5.0 410 .11E+10 6.781 .538 223 5.5 625 .14E+10 6.527 .625 270 6.0 673 .18E+1O 6.191 .506 .331 6.5 .863 .23E+1O 5.752 .478 .411 7.0 1.105 .29£+10 5.182 .441 516 7.6 1.415 .37E+10 4.446 391 650 8.0 1.812 .486+10 3.500 326 .823 8.6 2.316 .62E+10 2.293 241 1.044 9.0 2.949 .786+10 785 135 1.320 9.5 3.406 .91E+10 019 086 1.460 10.0 3.642 .97E+10 017 094 1.461 10.5 3.876 .1OE+11 016 103 1.461 11.0 4.107 .11E+11 014 111 1 461 11.5 4.335 .12E+11 013 119 1.462 12.0 4.560 .12E+11 .012 127 1 462 12.5 4.783 .13E+11 .011 134 1.462 13.0 5.002 .13E+11 .011 142 1.462 13.6 5.219 .14E+11 .010 149 1.462 14.0 6.433 .14E+11 .009 156 1.462 14.5 5.645 .156+11 .009 163 1.462 15.0 5.853 .1GE+11 .008 170 1.462 15.5 6.060 .16E+11 .008 176 1.463 16.0 6.263 .17E+11 .007 183 1 463 16.5 6.464 .17E+11 .007 189 1.463 1:38 Table A6 cont'd . EXPERIMENT 3 TIME xr NF SF AF PF (HR) (G/L) (CELLS/ML) (G/L) (G/L) (G/L) BATCH GROWTH 0.0 .029 .60E+08 8.200 .800 .014 .5 .037 .82E+08 8.171 .798 .019 1.0 .047 .11E+09 8.134 .795 .026 1.5 .060 .14E+09 8.086 .792 .035 2.0 .077 .19E+09 8.025 .787 .046 2.5 .099 .25E+O9 7.946 .782 .061 3.0 .127 .325+09 7.845 .774 .079 3.5 .163 .42E*09 7.716 .765 .103 4.0 .209 .54E+O9 7.551 .753 .133 4.5 .268 .7OE+09 7.339 .738 .172 5.0 .343 .90E+09 7.067 .718 .221 --5.3-START MICROFILTRATION - RESIDENCE TIME. 2.94 HR-— 5.5 .440 .12E+10 6.799 .691 .267 6.0 .563 .15E+10 6.575 .659 .301 6.5 .722 .19E+10 6.270 .623 .350 7.0 .925 .256+10 5.865 .582 .419 7.5 1.185 .32E+10 5.334 .534 .512 8.0 1.518 .40E+10 4.644 .476 .634 8.5 1.944 .525+10 3.754 .405 .794 9.0 2.486 .66E+10 2.614 .316 1.000 9.5 3.171 .856+10 1.177 .208 1.261 10.0 3.836 .10E+11 .029 .123 1.469 10.5 4.182 .11E+11 .025 .127 1.468 11.0 4.524 .12E+11 .021 .130 1.467 11.5 4.861 .13E+11 .019 .134 1.467 12.0 5.195 .14E+11 .017 .139 1.466 12.5 5.524 .15E+11 .016 .144 1.466 13.0. 5.849 .16E+11 .014 .149 1.465 13.5 6.170 .16E+11 .013 .154 1.465 14.0 6.487 .17E+11 .012 .160 1.464 14.5 6.800 .18E+11 .011 .165 1.464 15.0 7.109 .19E+11 .010 .171 1.464 15.5 7.414 .20E+11 .010 .176 1.464 16.0 7.715 .21E+11 .009 .182 1.464 16.5 8.013 .21E+11 .009 .187 1.464 17.0 8.307 .22E+11 .008 .193 1.463 17.5 8.598 .23E+11 .008 .199 1.463 18.0 8.884 .24E+11 .007 .204 1.463 18.5 9.168 .24E+11 .007 .210 1.463 19.0 9.447 .25E+11 .007 .215 1.463 19.5 9.723 .26E+11 .006 .221 1.463 20.0 9.996 .27E+11 .006 .226 1.463 20.5 10.266 .27E+11 .006 .232 1.463 139 Table A6 (cont'd) EXPERIMENT 4 TIME xr NF SF AF pr (HR) (G/L) (CELLS/ML) (G/L) (G/L) (G/L) BATCH GROWTH 0.0 039 .72E+08 8.200 .630 0.000 .5 049 .1OE+09 8.161 527 .007 1.0 063 .14E+09 8.111 524 016 1.5 081 .19E+09 8.046 519 028 2.0 104 .26E+09 7.964 513 043 2.5 133 .33E+09 7.858 505 063 3.0 171 .43E+09 7.723 496 087 3.5 219 .566+09 7.549 483 119 4.0 281 .72E+09 7.327 467 160 4.6 360 .93E+09 7.041 .446 212 6.0 .462 .12E+10 6.676 .420 279 ——-5.1—-START MICROFILTRATION - RESIDENCE TIME= 1.68 HR-— 5.5 .592 .16E+10 8.447 .607 288 6.0 .769 .2OE+1O 9.977 .770 305 6.5 .973 .2GE+1O 10.945 .878 344 7.0 1.248 .336+10 11.462 .943 407 7.5 1.601 .42E+10 11 596 .974 497 8.0 2.053 .556+10 11.385 .973 620 8.5 2.633 .7OE+1O 10.834 .944 782 9.0 3.377 .90E+10 9.923 .886 .993 9.5 4.330 .12E+11 8.608 .797 1.266 10.0 5.561 .15E+11 6.818 .672 1.617 10.6 7.112 .19E+11 4.456 .505 2.066 11.0 9.091 .24E+11 1.436 .289 2.632 11.5 10.813 .29E+11 049 .200 2.894 12.0 12.145 .32E+11 039 .214 2.903 12.5 13.460 I365+11 032 228 2.909 13.0 14.758 .39E+11 027 242 2.914 13.5 16.040 .43E+11 023 256 2.917 14.0 17.306 .46E+11 021 270 2.919 14.5 18.557 .506+11 018 283 2.921 15.0 19.791 .53E+11 016 296 2.922 15.5 21.011 .56E+11 015 309 2.923 16.0 22.215 .59E+11 014 322 2.924 16.5 23.405 .63E+11 013 335 2.925 17.0 24.580 .6eE+11 .012 348 2.925 17.5 25.740 .69E+11 .011 360 2.925 18.0 26.886 .72E+11 010 372 2.926 18.5 28.017 .7sE+11 .010 384 2.926 19.0 29.136 .786+11 .009 396 2.926 140 Table A6 (cont'd). 3 3 1 0 :a n 7; 0 pr. o rt .5 s V: c R U F 3‘.- ” 9h.0nh r62 0 0 I 71.0.... 0L2 L T. .L 0 0:0 0 E7 0 0.. Mn 2 ~400Mb " L O I.» I / 602 16 E 3 5365K .. 1.21.60 T. . u. u 3.1.. .93P 4. 01.1. 4. 0 o 0 0 .....1.1L "L .I L 1.1. $013.13 OF.» 0 OF .6 9 0X IX LNEX N 1% o ESA TIA» 61008 I 0 9 3~U H1. ANN-O IfiuH L 2 IQ BN 7. L .“ Nolll $535 5 0 [AE .. a CI 3 Duo :=.=.. :1 1 21.-U2 0 09.6 0 7.002 UUUZO D .6 QALQ EH79 TErur. LLLG3 S o /BAO SOC UX nu AAA—K0 L R. 4.. 4. . U03 05.7.; a VVV u .. 19.. .. . E U... cl.” 6 A .J 55 J 66394 .8 Q 3 .JJ . .3318 o o .1): I. 8 3 R...L.nu ML. 8 huh. .1 7 C~JHCiI 8 0063.6 9.002 2 HOXAV. OEE . ”N I 1555......16 05.1.2 6 UCflLhu<.q§HLI 7. 7. 1. ..v 0 III... 0 9.. M..3Mu.lp: Afflx 5.. wintry r0461 .1. 4......5 3.34.11.38VMP D: It .bPD 0 x 6 272 E 1. o 9.. ND. 5 AD. .3 2 “.1953 IL 204. A L CrLal 4. Hunk... in. O O OIL il .36.. “61,... .I o .311. ....H...rc Q. 151 n. o 361?.» 0 N «u? 2 o 1 PA U .CE7 [C 52430.0 U .1517“. HP 4. JJ~48 0:. 072.86 .Jp.29...31 3 9.» OCG 1.5.6.... o...§ EUu 6L»... Co. UBEEA 65 06.1... 5.12.313 ESE .6 369.1“ .969 0. 30. C3 C11... SSSLU .M. u LTK. 9...... P 64L UUUA. . 5.. CUtuuupL .N .CE X D. v... .. BOPAUOPPT ppGx APPPPHOU 3IJGLRJRCF RCLE HR CPCT. 8oooooooooooooooooooooooo00 [58899999911.1167339033333331 2n.UbUUrUUnb“2222221111111111R. oococooooooooooooooooooooo0 ’2229.22222226.22233333333312311. 100860000080800000000000008 10000000000000...0000000000 5 5555555555555555555555555 11111111111111111111111111 LITERATURE CITED LITERATURE CITED Aiba, S., Shoda, M., and Nagatani, M., "Kinetics of Product Inhibition in Alcohol Fermentation," Biotechnology and Bioengineering, 12, 845 (1968). Amarasingham, C. R. and Davis, B. D., "Regulation of d—KetOglutarate DehydrOgenase Formation of Escherichia caZi," Journal of Biological Chemistry, 240, 3664 (1965). Andrews, J. F., "A Mathematical Model for the Continuous Culture of Microorganisms Utilizing Inhibiting Sub- strates," Biotechnology and Bioengineering, 12, 707 (1968). Bailey, J. B., and D. F. Ollis. Biochemical Engineering Fundamentals. McGraw-Hill: 1977. Beck, C., and von Meyenberg, H. K., "Enzyme Pattern and Aerobic Growth of Saccharomyces cerevisiae Under Various Degrees of Glucose Limitation," Journal of Bacteriology, 96, 479 (1968). Bijkerk, H. E. and Hall, R. J., "A Mechanistic Model of Aerobic Growth of Saccharomyces cerevisiae," Biotechnology and Bioengineering, 24, 267 (1977). Bolivar, F., Rodriques, R. L., Betlach, M. C., and Boyer, H. W., "Construction and Characterization of New Cloning Vehicles I. Ampicillin—Resistance Derivatives of the Plasmid pMB9," Gene, 2, 75 (1977a). Bolivar, F., Rodriquez, R. L., Greene, P. J. Betlach, M. C., Heyneker, H. L., Boyer, H. W., Crosa, J. H., and Falkow, S., "Construction and Characterization of New Cloning Vehicles II. A Multipurpose Cloning System," Gene, g, 95 (1977b). Bolivar, F., and Backman, K., "Plasmids of Escherichia caZi as Cloning Vectors," Methods in Enzymology, fig, 245 (1979). Boyer, H. W. and Roulland-Dussoix, D., "A Complementation Analysis of the Restriction and Modification of DNA in Escherichia coli," Journal of Molecular Biology, 41, 459 (1969). Charley, R. C., Fein, J. C., Lavers, B. H., Lawford, H. G., and Lawford, G. L., "Optimization of Process Design for Continuous Ethanol Production in Zymomonas mobilis ATCC 29191," Biotechnology Letters, 5, 169 (1983). 141 142 Chesney, A. M., "The Latent Period in the Growth of Bacteria," J. of Exp. Med., 24, 387 (1916). Coulman, G. A., Stieber, R. W., and Gerhardt, P., "Dialysis Continuous Process for Ammonium-Lactate Fermentation of Whey: Mathematical Model and Computer Simulation," Applied and Environmental Microbiolggy, 34, 725 (1977). Doelle, H. W., Ewings, K. N., and Hollywood, N. W., "Regulation of Glucose Metabolism in Bacterial Systems," Advances in Biochemical Engineering, 23, l (1982). Doelle, H. W., Hollywood, N., and Westwood, A. W., "Effect of glucose concentration in a number of enzymes involved in the aerobic and anaerobic utilization of glucose in turbidostat cultures of Escherichia coZi," Microbios, g, 221 (1974). Dostalek, M. and Haggstrom, M., A Filter—Fermentor-Apparatus and Control Equipment," Biotechnology and Bioengineering, 24, 2077 (1982). Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A., and Smith, F., "Coloriometric Method for Determination of Sugars and Related Substances," Analytical Chemistry, 28, 350 (1955). Endo, I., Nagamune, F., and Inoue, I., "Start—up of the Chem- ostat Culture," Annals of the New York Academy of Sciences, 369, 245 (1981). Faber, M., "Meeting Report: Cornell Scientists Clone, Express Cellulase Genes in E. coli, Biotechnology, l, 139 (1983). Franson, M. A. (ed). Standard Methods for the Examination of Water and Wastewater. 14th edition. American Public Health Association: 1976. Gaden, E. L., "Fermentation Process Kinetics," Journal of Biochemical and Microbiological Technology and Engineering, I, 413 (1959). Gerhardt, P., R. G. E. Murphy, R. N. Costilow, E. W. Nester, W. A. Wood, N. R. Krieg, G. B. Phillips. Manual of Methods for General Bacteriology. American Society for Microbiology: 1981. Henry, J. D. and Allred, R. C., "Concentration of Bacterial Cells by Cross Flow Filtration," Developments in Industrial Microbiology, 13, 177 (1972). Herbert, D., Elsworth, R., and Telling, R. C., "The Continuous Culture of Bacteria; a Theoretical and Experimental Study," Journal of General Microbiology, 14, 601 (1956). 143 Hollywood, N., and Doelle, H. W., "Effect of specific growth rate and glucose concentration on growth and glucose metabolism of Escherichia coli K—12," Microbios, 11, 23 (1976). Holzberg, I., Finn, R. K., and Steinkraus, K. H., "A Kinetic Study of the Alcohol Fermentation of Grape Juice," Biotechnology and Bioengineering, g, 413 (1967). Ingraham, J. L., O. Maaloe, F. C. Neidhardt. Growth of the ' Bacterial Cell. Sihauer Associates, Inc}: 1983. Johnson, M., "Isolation and Properties of a Pure Yeast Poly— Peptidase," Journal of Biological Chemistry, 137, 575 (1941). Johnson, R. R., Balwani, T. L., Johnson, L. J., McClure, K. B., and Dehority, B. A., "Corn Plant Maturity. II. Effect on in vitro Cellulose Digestability and Soluble Car— bohydrate Content," J. of Animal Science, 2;, 617 (1956). Kossen, N. W. F., "Mathematical Modelling of Fermentation Processes: Scope and Limitations," Symp. Soc. Gen. Microbiology, 29, 327 (1979). Landwall, P. and Holme, F., "Removal of Inhibitors by Dialy— sis Culture," Journal of General Microbiology, 103, 345 (1977). Levenspiel, 0., "The Monod Equation: A Revisit and a Gen- eralization to Product Inhibition Situations," Biotech— nology and Bioengineering, 22, 1671 (1980). Lodge, R. M. and Hinshelwood, C. N., "The Lag Phase of Bact. Lactis Aerogenes," J. Chem. Soc., 213 (1943). Luedeking, R. and Piret, E. L., "A Kinetic Study of the Lactic Acid Fermentation. Batch Process at Controlled pH," Journal of Biochemical and Microbiological Tech- nology and Engineering, 1, 393 (1959). Marr, A. G., Nilson, E. H., and Clark, D. J., "The Maintenance Requirement of Escherichia coZi," Annuls of the New York Academy of Science, 122, 536 (1963). Mateles, R. I., Ryu, D. Y., and Yasuda, T., "Measurement of Unsteady State Growth of Micro-organisms," Nature, 208, 263 (1965). Monod, J., "The Growth of Bacterial Cultures," Ann. Review of Microbiology, 3, 371 (1949). 144 Montgomery, R., "Further studies of the phenol-sulfuric acid reagent for carbohydrates," Biochim. BiOphys. Acta, 42, 591 (1961). Pamment, N. B. and Hall, R. J., "Absence of External Causes of Lag in Saccharomyces cerevisiae," J. of Gen. Micro- biology, 105, 297 (1978). Pamment, N. B., Hall, R. J., and Barford, J. P., "Mathematical Modeling of Lag Phases in Microbial Growth," Biotech— nology and Bioengineering, 29, 349 (1978). Penfold, W. J., "On the Nature of Bacterial Lag," J. of Hygiene, 24, 215 (1914). Peringer, P., Blachere, H., Corrieu, G., and Lane, A. G., "Mathematical Model of the Kinetics of Growth of Saccharomyces cerevisiae," Biotechnology and Bioengineer- ing Symposium, 4, 27 (1973). Peringer, P., Blachere, H., Corrieu, G., and Lane, A. G., "A Generalized Mathematical Model for the Growth Kinetics of Saccharomyces cerevisiae with Experimental Deter- mination of Parameters," Biotechnology and Bioengineering, 26, 431 (1974). Pirt, S. J., and Kurowski, W. M., "An Extension of the Theory of the Chemostat with Feedback of Organisms. Its Experimental Realization with a Yeast Culture," Journal of General Microbiology, 62, 357 (1970). Repaske, R., Repaske, A. C., and Mayer, R. D., "Carbon Dioxide Control of Lag Period and Growth of Streptococcus sanguis," J. of Bacteriology, 117, 652 (1974). Rogers, P. L., Lu, K. J., Shotnicki, M. L., and Tribe, D. B., "Ethanol Production by Zymomonas mobilis," Advances in Biochemical Engineering, 22, 37 (1982). Sortland, L. D. and Wilke, C. R., "Growth of Streptococcus faecalus in Dense Culture," Biotechnolggy_and Bio- engineering, 21, 805 (1969). Stanier, B. Y., E. A. Adelberg, J. Ingraham, The Microbial World, 4th edition, Prentice-Hall: 1976. Stieber, R. W., and Gerhardt, P., "Continuous Process for Ammonium-lactate Fermentation of Deproteinized Whey," Journal of Dairy Science, 62, 1558 (1979). Supelco, Inc., "Extraction Procedures for GC Analysis of Culture By-products for Volatile Acids and Alcohols," Supplement to Bulletin 7486, 1975. 145 Swanson, C. H., Aris, R., Fredrickson, A. G., and Tsuchija, H. M., "Bacterial Growth as an Optimal Process," J. of Theo. Biology, 22, 220 (1966). Tanny, G. B., Merilman, D., and Pistole, T., "Improved Fil- tration Technique for Concentrating and Harvesting Bacteria," Applied and Environmental Microbiology, fig, 269 (1980). Watson, D. C., and Berry, D. R., "Use of an Exchange Fil- tration Technique to Obtain Synchronous Sporulation in an Extended Batch Fermentation, "Biotechnology and Bio- engineering, 22, 213 (1979). Yang, R. D., and Humphrey, A. E., "Dynamic and Steady State Studies of Phenol Biodegradation in Pure and Mixed Cultures," Biotechnology and Bioengineering, 21, 1211 (1975). MATERIAL BIBLIOGRAPHY 1M 2M 3M 4M 5M 6M 7M 8M 9M 10M 11M 12M 13M 14M 15M Material Bibliography American Instrument Company, Inc. Silver Springs, Maryland Beckman Instruments Inc. Fullerton, California Cole-Parmer Instrument Co. Chicago, Illinois Difco Laboratories Detroit, Michigan Drexelbrook Engineering Co. Horsham, Pennsylvania Gelman Sciences, Inc. Ann Arbor, Michigan Leeds and Northrup North Wales, Pennsylvania Leeds and Northrup North Wales, Pennsylvania Linear Instrument Corporation Reno, Nevada New Brunswick Scientific Company New Brunswick, New Jersey New Brunswick Scientific Company New Brunswick, New Jersey Sigma Chemical Company St. Louis, Missouri Sigmamotor Middleport, New York Supelco, Inc. Bellefonte, Pennsylvania Supelco, Inc. Bellefonte, Pennsylvania 146 Peristaltic pump SpectrOphotometer Model DB-G Masterflex Variable Speed System Cat. No. 7549-19 Plate Count Agar Level Transmitter Model YC 508-25-2 Acroflux Capsule Autoclavable pH electrodes Current controller Centry Model 440- —l-20—33-51-A104 Chart Recorder Model 141 Microferm fermentor pH Controller Model pH-4O Antifoam A concentrate Peristaltic pump Model T-GS Volatile Acid Standard Mix Column Packing SP-lOOO on 100/120 Chromosorb W Aw A 16M 17M 18M 19M 147 Supelco, Inc. Bellefonte, Pennsylvania Supelco, Inc. Bellefonte, Pennsylvania Varian Associates Walnut Creek, California Varian Associates Walnut Creek, California Non-volatile Acid Mix Column Packing 15% DEGS on 80/100 Chromosorb W AW Gas Chromatograph Series 1400 Integrator CDS-lll M'TIWITIEIWLTIfl11111fllflfllfltflhflflfliflifllflllmS