LlOUID a LIQUED EXTRACTION [N A ' PULSEB QOLUMN Thesis for “19 Degree of DH. D. MICHIGAN STATE UNIVERSITY Clayton Daie Caiiihan 1957 This is to certify that the thesis entitled LIQUID-LIQUID EXTRACTION IN A PULSED COLUMN presented bg CLAYTON DALE CALL IHAN has been accepted towards fulfillment of the requirements for Doctor of Philosophy degree in Chemical Engineering Major professon/ Date May 17: 1957 0-169 - é a ‘. LIQUID-LIQUID EXTRACTION IN A PULSED COLUMN BY CLAYTON DALE CALLINAN Submitted to the School for Advanced Graduate Studies of Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemical Engineering 1957 Copyright by Clayton Dale Callihan 19 59 Clayton Dale Callihan ABSTRACT The performance of a packed liquid—liquid extraction column may be improved if a pulsating motion is imparted to the liquid in the column. The height equivalent to a transfer stage (HETS) is reduced when either the pulse amplitude or pulse frequency are increased. At low pulse rates the amplitude-frequency product seems to be the determining factor, but at high rates frequency is more beneficial than amplitude. The equivalent number of transfer 8 ages in a given height was l/lh as many in an unpulsed column as in a pulsed column when an amplitude of 5 mm and a frequency of 215 cycles per minute were used. HETS was found to be more useful than HTU in evaluating the performance of these columns, since the latter varied strongly with flow ratio. HETS is largely a function of packing characteristics, pulsation rate, and superficial throughput velocity. Increasing the cross-section of the packed column by a factor of 2.h3 (from 2.127-inch to 3. 32-inch ID) did not significantly change the HETS if the superficial velocity, pulse amplitude, and pulse frequency were held constant. Settling and reorientation of the packing as a result of pulsation had an appreciable effect on NETS for both pulsed and unpulsed operation. In studying the influence of operating variables on the maximum throughput velocity (flooding velocity), one variable was found unexpectedly to dominate the results. This was the rate of mass transfer of the solute from one phase to the other. In a section -11- Clayton Dale Callihan 3l inches high packed with 8-mm Raschig rings and using carbon tetrachloride and water as solvents, the flooding velocity was twice as high when the entering CClu contained 1% acetone than it was when no acetone was present in either phase or when the acetone concentration in both phases was in equilibrium. Increasing the column height to 101 inches reduced the mass transfer per unit height and therefore reduced the flooding velocities. Increasing the ratio of water to CClu from O.h to h.0 caused a composition "pinch" at the bottom of the column, and the lack of mass transfer at this point reduced the flooding velocity 50%. The effect of pulsing was to increase the average mass transfer per unit of height, although at some flow ratios the pinching effect also became more severe. Pulsing increased the flooding velocity in some cases and decreased it in others. Experimentally this large effect of mass transfer on flooding rate made it difficult to measure the much smaller influence of other variables. From a design viewpoint, it casts doubt on the practical usefulness of the correlations of Hoffing and Lockhart, Breckenfield and Wilke, and other who Obtained all their data in the absence of a solute. - iii - TABLE OF CONTENTS PAGE ABSTRACm . . . . . . ACWO‘JIJEDGEIVENT o o o o o 0 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Classification of Extractors. . . . . . . . . . . . . . . . . . 2 Miscellaneous Types. . . . . . . . . . . . . . . . . . . . h Mixer-Settler Types. . . . . . . . . . . . . . . . . . . . h Pulsed Columns . . . . . . . . . . . . . . . . . . . . . . 5 Previous Work on Pulsed Columns . . . . . . . . . . . . . . . . 6 Sieve-Plate Columns. . . . . . . . . . . . . . . . . . . . 6 Spray Columns. . . . . . . . . . . . . . . . . . . . . . . ll Pulsed Spray Columns . . . . . Packed Columns . . . . . . o o o o o o o o o o o I—J r0 0 C O O O O O O O I O O O O O O 12 Pulsed-Packed Columns. . . . . . . . . . . . . . . . . . . 1h mmose O O O O O O O O O O O O O O O O O O O O O O O I O O O O 21 Limitations and Scope. . . . . . . . . . . . . . . . . . . 21 APPARArj-US mm PRmEDm O O O O O O O O O I O O O O O O O O O O O O 0 21+ Solvent-Solute Systems . . . . . . . . . . . . . . . . . . 31 Analytical Procedures. . . . . . . . . . . . . . . . . . . 31 Operating Procedures . . . . . . . . . . . . . . . . . . . 3h Accuracy and Reproducibility . . . . . . . . . . . . . . . 3 P‘IethOd Of CdC‘lJ—ation O O O O O O I 0 O O Q 0 O O O O O O 0 1+1 EMILE ITAL RESLJLTS o o o o o o o o o o o o o o o o o o o o o o o 0 M6 Distribution of Acetone in Water and CClu. Small Column Operation. . . . . . . . . . . . . . . . . . . . . AB Unpulsed Runs on Loosely Settled Packing . . . . . . . . . M8 Pulsed and Unpulsed Runs on Settled Packing. . . . . . . . AB Haring Blender Tests . . . . . . . . . . A6 . . . . . . . . . . . . . . . . . 51 Pulsed and Unpulsed Runs Containing No Acetone . . . . . . 52 Pulsed and Unpulsed Runs Containing Acetone. . . . . . . . 55 Pulsed Runs Utilizing Only Part of the Column Capacity . . . . . . . . . . . . . . . . . . . . . . . . 55 Large Column Operation. . . . . . . . . . . . . . . . . . . . . 59 Expanded End Sections. . . . . . . . . . . . . . . . . . . 6O Unpulsed Runs on Loosely Settled Packing . . . . . . . . . 61 Unpulsed Runs on Well Settled Packing. . . . . . . . . . . 61 Pulsed Runs on Well Settled Packing. . . . . . . . . . . . 65 -iv- TABLE OF commas (Continued) Page DISCUSSIONOFRESULTS.................... 66 Factors Which Influence Flooding Rates. . . . . . . . . 66 Height of Adjustable Overflow Leg. . . . . . . . . 66 Flow Ratio . . . . . . . . . . . . . . . . . . . . 67 Effect of Solute Transfer. . . . . . . . . . . . . 68 Physical Properties of the Liquids . . . . . . . . 72 Packing Height . . . . . . . . . . . . . . . . . . 7h Packing Density. . . . . . . . . . . . . . . . . . 77 Pulse Amplitude. . . . . . . . . . . . . . . . . . 80 Pulse Frequency. . . . . . . . . . . . . . . . . . 88 Particle Dispersion. . . . . . . . . . . . . . . . 92 Factors Which Influence Column Efficiencies . . . . . . 93 Throughput Rates . . . . . . . . . . . . . . . . . 93 Flow Ratio . . . . . . . . . . . . . . . . . . . . 99 End Effects. . . . . . . . . . . . . . . . . . . . 106 Orientation of Packing . . . . . . . . . . . . . . 107 Packing Density. . . . . . . . . . . . . . . . . . 109 Pulse Amplitude. . . . . . . . . . . . . . . . . . 112 Pulse Frequency. . . . . . . . . . . . . . . . . . 116 Miscellaneous. . . . . . . . . . . . . . . . . . . 118 Comparison of the Two Columns . . . . . . . . . . . . . 121 Unpulsed Packed Columns. . . . . . . . . . . . . . 121 Pulsed Columns . . . . . . . . . . . . . . . . . . 123 Comparison with Other Columns. . . . . . . . . . . 12h COIJCLUSIONS O O O O O O O O O O O O O O O O O O O O O O O O O 125 APquDH I O O O O O O O O O O O O O O O O O O O O O O O O O 129 BIBIIIOGMPIH O O O O O O O O O O O O O O O O O I O O O O O 0 1'30 TABLE NO II III IV VI VII VIII IX XI XII XIII XIV XVI XVII XVIII XIX LIST OF TABLES Data on Pulsed Sieve-Plate Column - Sege and Woodfield . . Data on Pulsed Sieve-Plate Column - Chantry et a1. . . . . Data on Pulsed Sieve-Plate Column - Cohen and Beyer. . . . Data on Pulsed Spray Column - Billerbeck et al . . . . . . Data on Pulsed-Packed Columns - Feick and Anderson . . . . Data on Pulsed-Packed Columns - Schuler. . . . . . . . . . Data on Pulsed-Packed Columns - Chantry et a1. . . . . . . Physical Properties on Liquids used in Tables I through VII 0 O O O O O 0 O 0 O O O O O O O O O O O O O O O O 0 Dependence of Accuracy on Exit Raffinate Concentrations. . Distribution of Acetone in Water and Carbon Tetrachloride. Distribution of Acetone in water and Carbon Tetrachloride (corrected) 0 O O O O O O O O O O O O O O O O O O O O O O Unpulsed Runs on Loosely Settled Packing - 2.062h-inch C Olumn . O C C . O O O O C O . O O O C C Q C C C O O O . Pulsed and Unpulsed Runs on Loosely Settled Packing - 20127‘inCh (3011117111. 0 o o o o o o o o o o o o o o o o o o waring Blender Tests for Emulsification. . . . . . . . . . Pulsed and Unpulsed Runs with No Acetone Present - 20127‘inCh COlUIm. o o o o o o o o o o o o o o o o o o o Pulsed and Unpulsed Runs on Well Settled Packing — 2ol27‘inCh COlum. o o o o o o o o o o o o o o o o o o o Pulsed Runs Using Only Part of the Column Capacity . . . . Unpulsed Runs on Loosely Settled Packing - 3.32-inch COlumn O O O O O O O 0 O O O O O O O O O O O O Unpulsed Runs on Well Settled Packing - 3.32-inch Column . Unpulsed Runs on Well Settled Packing - 3.32-inch Column . -vi- A6 A7 53, >4 62 63 6h, 6AA TABLE NO XXI XXII XXIII XXIII(a) XXIV XXVI XXVII XXVIII XXXI XXXII XXXIII LIST OF TABLES (Continued) Unpulsed Flooding Rates on 2.062b-inch Column . . . . . . Comparison of the Flooding Rates of the Two Columns . . . Effect of Column. Effect of Influence Column. Influence Influence Effect of Effect of Effect of Column. Effect of Column. Effect of Column. Effect of Reduced Throughput on HETS (Unpulsed) 2.06-inch Reduced Throughput (Pulsed at 125 RPM). . . . . of Flow Ratio on HETS and BTU - 2.062-inch of Flow Ratio on HETS and HTU . . . . . . . . . of Flow Ratio on HETS and ETU . . . . . . . . . Packing Orientation on HETS . . . . . . . . . . Packing Density on HETS at Flooding (Unpulsed). Amplitude on HETS at Flooding - 2.127-inch Amplitude on HETS at Flooding 3.32-inch Frequency on HETS at Flooding - 2.127-inch Column Diameter on HETS Values. . . . . . . . . Comparison of HETS Values (Pulsed). . . . . . . . . . . . - vii - PAGE 101 101 10h 108 110 112 116 122 12h LIST OF FIGURES FIGURE N0 PAGE 1 Schematic Drawing of a Pulsed-Packed Column . . . . . . . . 25 2 Photograph of the 2.127-inch Column . . . . . . . . . . . . 26 3 Photograph Of the 3.27-11’101‘1 COlunmo o o o o o o o o o o o 0 26A h Dependence of Accuracy on Final Raffinate Concentration . . A2 5 Graphical Methods for Determining the Number of Theoretical Stages. 0 O O O O O I O O O O O O O O I O O I O O O O O O 1‘5 6 Unpulsed Flooding Rates - 2.127-inch Column . . . . . . . . 69 7 Runs with Small Amount of Acetone Present - 2.127-inch COlumn. C O C C O O O O O O O O O O O O O O O I C O I O O 71 8 Unpulsed Flooding Runs on the 3.32-inch Column. . . . . . . 75 9 Unpulsed Flooding Runs on the 3.32-inch Column. . . . . . . 76 10 Affect of Packing Density on Throughput Rates . . . . . . . 79 11 Comparison of Flooding Rates of the Two Columns . . . . . . 81 12 Effect of Amplitude on Flooding Rates - 2.127-inch Column . 83 13 Effect of Amplitude on Flooding Velocities at Two Flow Ratios. . O O O O C O O O O O I O O O O O O O O O O O O O 81+ 1h Pulsed and Unpulsed Flooding Rates with 1% Acetone in , organic Pmse O O O O O O O O O O I O O O O O O O O C C 0 80 15 Effect of Frequency on Flooding Rates - 2.127-inch Column . 89 16 Effect of Frequency on Flooding Velocities at Various Flow Ratios. . O O O O O I O O O O O O I O O O O O O O C O O 0 9:1- 17 Effect of Reduced Throughput on HETS - 2.062h-inch Column . 95 18 Effect of Reduced Throughput on 11ers (Pulsed) — 2.127-inch COluUm. . O O O O O O O O O O I C C O O O O O O C C . O . 98 19 Comparison of HTU Values with HETS Values . . . . . . . . . 102 - viii - LIST OF FIGURES (Continued) FIGURE N0 PAGE 20 Influence of Flow Ratio on HETS and HTU at Flooding . . . . 103 21 Graphs Showing the Variation of HTU and HETS with Flow Ratio 0 O O O O O O I O O I O O O O O O O O O O O O .0 O O lOS 22 Effect of Packing Density on HETS at Flooding (Unpulsed). . 111 23 Effect of Amplitude on HETS at Flooding - 2.127-inch Column 113 2h Effect of Amplitude on NETS at Flooding - 3.32-inch Column 115 25 Effect of Frequency on HETS at Flooding - 2.127-inch Column 117 26 Effect of Pulsed Volume on HETS at Flooding - 2.127-inch COlumn O O O O O O O O O O I O O I C O O O O O I O O O O O J—lg 27 Comparison of HETS Values for the Two Columns (Pulsed). . . 12AA -ix- APPRECIATION The author wishes to express his sincere appreciation to Dr. Carl M. Cooper for the many helpful suggestions and the generous use of his time for consul throughout the course of this investigation. Thanks are also due to William B. Clippinger for constructing the mechanical equipment necessary for this research. Appreciation is also extended to the Dow Chemical Company for supplying a scholarship which covered part of the expense of this investigation. INTRODUCTION Continuous liquid-liquid extraction has been used industrially to great advantage as a unit operation in the separation and purification of chemicals. This is primarily due to certain basic features inherent to the liquid-liquid extraction process itself. One point in favor of extraction is that no heat or steam is required for the extraction step. Some heat may, however, be required to separate the product from the solvent. In several processes, such as the separation of acetic acid from water solutions, liquid-liquid extraction has contrib- uted greatly to the economics of the process. In the separation of high molecular weight compounds or heat sensitive materials, distillation is sometimes impractical because the materials can- not be vaporized without the use of very high vacuum and extraction is the only reasonable separation method. Although similar in principle to fractional distillation, liquid- 1iquid extraction has an advantage in that the solvents can be chosen from thousands of compounds available commercially to give a great preference for one or more of the components. About the only restrictions placed on the solvents are that they must be relatively insoluble in each other. It is readily apparent, therefore, that the selectivity of the two liquids can often be made greater than in distillation where the vapor phase is substantially ideal and the activity of each component in the vapor is very nearly proportional to its concentration. Liquid-liquid extraction can, therefore, be made to give a greater degree of separation in a single equilibrium contact than is obtainable in fractional distillation. - 2 - Unfortunately, the height equivalent to an equilibrium contact in an extraction column is generally much greater than the height equiv- alent to a theoretical contact in a distillation column, and this has often limited the use of extraction. Thus if liquid-liquid contactors could be confidently designed with nearly the same stage height as a distillation column, a contribution would be made to the chemical industry. The present investigation is one contribution toward solving this inefficiency problem, and presents data on extraction columns which in some tests showed even better stage efficiencies than distillation columns. Classification of Extractors Liquid-liquid extractors have been classified in an exhaustive study made by Morello and Poffenberger (l). The two main classifications differ depending on whether gravity or centrifugal force is used to separate the phases. Most extractors used industrially are of the gravity type, and they can be further subdivided by whether the contact is made through extended films of the two phases or through dispersed droplets of at least one of the phases. Those extractors which depend on droplet formation for operation may be further sub- divided into those using power to maintain drop dispersion, and those that do not. Morello and Poffenberger pointed out that a preference was shown in industrial designs for those extractors which do not use power. Extractors that use power result in added costs, not only for the power consumed but also for the cost of maintenance of shafts, stuffing boxes, and other moving parts. Nearly all of the extractors which used an outside source of power were essentially mixer and - 3 - settler contactors employing a variety of flow patterns and arrange- ments. Some of the main objections in industrial application to those extractors which use an outside source of power have been: 1. Inefficient use of the power being supplied to the extractor. 2. Too little knowledge of reliable design methods for this equipment. 3. Lack of a necessary motive for changing the design of equipment which is now performing quite well its intended function. h. Extra cost involved when columns are shut down for repair of the mechanical parts. Pulse columns do not have this handicap because the pulsator is situated externally. Despite these apparent disadvantages, it is recognized by a great many authors that extractors that use an outside source of power often require less space and less investment to make the same separations than their non-agitated counterparts. It should furthermore be pointed out that 1950, the year the Morello-Poffenberger article was published, also marked the acceleration of emphasis on contactors with an outside power source. This was primarily because of the rapid interest developed by the United States government in columns that could give a great many stages in a short height in order to reduce the cost of shielding for columns extracting radio-active materials. In view of this increased interest in extraction columns employing power, and particularly in those employing pulsation, power columns will be classified in this thesis as follows: 1. Pulsed columns. 2. Mixer-settler types of apparatus. 3. Miscellaneous power driven extractors. - h _ Miscellaneous Types The miscellaneous classification includes only a few special types which cannot be considered as falling in the first two categories. The most important of these is the Podbelniak spiral extractor which employs centrifugal force to cause the liquid films to flow countercurrently in contact with each other. Its cost has been justified only for special applications where low holdup time is particularly important. Mixer-Settler Types A.typical mixer-settler extractor consists of a multiplicity of chambers with alternate chambers equipped with mechanical agitators and the others arranged for settling and decanta- tion. A countercurrent flow pattern is used. Since true equilibrium is approached closely in each stage, such extractors may be calculated and designed with complete assurance. On the other hand, they are often expensive, complex, and bulky, requiring a large amount of floor space. Special designs have been introduced in an effort to minimize these undesirable features. Modified mixer-settler types represent compromises, and sacrifice efficiency per stage for a more convenient and compact arrangement. The Scheibel column is one such compromise. In the Scheibel column a series of small mixers are connected to a central rotating shaft running vertically through the column. Between the mixing blades are non-agitated sections of the column packed with fine wire mesh. These sections act as settling chambers or calming regions where the finely divided droplets formed in the agitated sections have a chance to coalesce. Scheibel columns some- times give more than one theoretical stage for each pair of agitators and separation sections because of countercurrent action in the wire mesh. Hewever, the efficiency depends on the system being extracted - 5 - and the column diameter. Data reported in the literature show that a minimum theoretical stage height of one foot can be expected on a lE-inch column with a maximum throughput rate of about half that of an ordinary packed column. Even pulsed columns sometimes include a sort of mixing-settling action. A pulsed sieve-plate column, for example, operating at low speeds, certainly has an area for dispersion, and the dispersed liquid is moved to another region for coalesence. However, all pulsed columns do not have these features and it is customary to classify them as a separate group. Pulsed Columns If an up-and-down motion is superimposed on the net countercurrent flow of the two phases going through an extraction column, the result is a pulsed column. Although such columns were first described in the literature in 1937, very little interest actually developed until about 1950. Almost any type of construction can be used inside the pulse column for performing the necessary dispersing and coalescing operations. For example, a series of sieve-plates could be used, in which the liquids are dispersed as they are forced through small perforations and allowed to coalesce in the region between the plates. Packing could also be used, in which the drops are dispersed on rapid contact with the stationary packing and allowed to coalesce in the spaces between. Spray columns have been tried as pulse columns, in which the dispersion is obtained by introducing a fine mist or spray. It has recently been pointed out to the author that baffle plate tOWers have been tried as pulsed columns, although these efforts have met with very little SUCCESS. - 6 - Previous Work on Pulsed Columns Most pulsed columns have been designed from standard unpulsed columns except that a pulsing feature has been added. This makes it impossible to discuss columns using an outside source of power with- out at the same time discussing the original column from which it was derived. This report will make no attempt in the following discussion to separate pulsed and unpulsed columns but will describe them together as the occasion arises. Sieve-Plate Columns The first mention of pulsed columns in the literature was a patent issued to w. J. D. van Dijck (2) in 1937. In this patent, van Dijck described two different types of pulsed columns. One of them consisted of a series of perforated plates, commonly called sieve-plates, placed one above the other in a vertical column. Unlike the usual perforated plate column, they contained no downcomers for the heavy phase. The sieve-plates were connected to each other and the top plate was fastened by a shaft to a motor- driven eccentric. The reciprocating motion of the eccentric caused all of the plates in the column to move up and down. To the writer's knowledge, few, if any, except eXperimental columns were ever built of this design. Another column mentioned in the van Dijck patent has found considerable popularity. This, the pulsed sieve-plate column is widely used in the atomic energy program in a variety of sizes. The sieve-plates remain stationary while an up-and-down motion is superimposed on the countercurrent flow of the two liquid phases by a pulsator which forces liquid in and out of the bottom of the column. If the holes in the plates are sufficiently small so that a high degree - 7 - of dispersion is obtained, then liquid cannot flow countercurrently through the column unless the pulse is operating. This fact offers certain advantages, since during a temporary shut down of the pulsator, unextracted liquid cannot get through the column. Numerous articles on pulsed sieve-plate columns have appeared in recent literature. Foremost among these is the report by Sege and Woodfield (3), who worked on the separation of uranyl nitrate. These men, operating both a three-inch and an eight-inch diameter column, investigated a great number of the variables involved in the operation of sieve-plate columns. Bhny of the results that seem typical of this type of contactor are recorded in Table 1. Of possible interest in special applications is a subsequent report (h) on a 23.5-inch column. This shows that where channeling is due to the liquid at the top of the column being heavier than that at the bottom, channeling may be prevented by means of a louver-plate redistributor. An excellent article prepared by Wiegandt and von Berg (5), presents some of the particular problems involved in the operation of both packed and sieve-plate pulsed columns. While this report does not present any actual data, a later report prepared by Chantry (6), for a doctoral thesis under the direction of these two men gives the results of runs on a 1.57-inch diameter pulsed sieve-plate column. Typical data from their report are tabulated in Table II. Cohen and Beyer (7), give results obtained on the performance of a one-inch diameter sieve-plate column, extracting boric acid from isoamyl alcohol into water. They Obtained values of a contact stage as low as 9.9 inches under certain pulse conditions. Some of these runs are tabulated in Table III. -- mm Hows. .sm s s.0 -- 0 -- oaaso. 0.0us 0s.0m H” -- onassso Hm : : H.H nu mmnmo. u: I: m.>m OH.O>H nu nu m50m5v¢ OC Haas . a.0 -- masmo. -- -- m.sa 0s.0»s -- -- guesses as : 2 @.H II mOJNOo I... II N.k_..n\v OH.ON.H II II mSOGHJRud me. ogwgpogosau : 0.0 -- m03m0. -- -- m.m0 00.0Ha - -- causes an : : OoH ll Amij. II II womdfl ONommm I I... mfiowgdd WH Hmopm .pm 0 0.H -- mommo. 0 0.ma 00.ema -- -- caesmso ma : : w.o u- u .. ommzo. H.m® Om.m: n“ I- mmomswd :a ocogponozaa ; v.0 I- u nu ommzo. 0;.0H on.mm I I: oasmmso ma : z 0.0 -- - -- 000:0. m.wsa 0H.0w -- -- . ma amass .pm a :.H -- 0 -- 000a0. 0.s0a 0m.mm H-\ 0.H e Ha : : ®.o u: mmjmo. I: In m.:m :A.N: 0.0x A. mSode< OH memnpososac . m.a -- mmsmo. -- -- m.:m sm.m: 0 00 L. 0 t Z NoH II mmimoa II II momw. 00oFNH 0.0m 00H 2 w z 2 MoH I... mOJNOo II II moNN OQ.\LNH 0.0“H 00H. 2 N. . : O H.H. II mOINNO. II I... m.N.~ OOoNLNH OomwOH 0. 2 O 0% z : No0 II 0 II OQQJO. OoPOH Omomm WHWM moH. : m _ = z 0.0 -- 0 -- 000:0. 0.s0a 0m.mm a m0 0. z a z = 0.0 -- 0 -- 000:0. 0.>0H om.mm 0.00 m z m . . s.0 -- 0 -- Aaio. 0:.ma ms.0m 0.0: m.H . = z 0.0 -- 0 -- 000:0. 0:.mm ms.mm m.m0 0. z m Hmmpm.pm s 0.0 -- 0 -- 000:0. 0:.mn ms.mm 0.00 m. onesmso a sass puma p20 cH p20 CH madam swam: madam pamfiq mazes: smasm emmhmmufia oz cam mamas as: swung a>soa madam sauna .pa ammsmxwa :0 hog mmeoeH mmanm Pm s©\mofio£ DA onoo mpSHom mam: moan oflnumszao> mmdomo H6908 QAMHmQOOB .3 .m 92¢ momm .0 20mm ¢H¢Q .:®OH ZOHBOHm mB¢Am m0 Bmon: .mMBdE 92¢ HBDAOm Mme md meamBHz qwz¢mb mBHZ mQHmOAmodemBzomm¢o a AO> ON Qz<.madmmmo:m ANBDmHmB a 40> Om OZHmD «dmmd Emma fimm UZH>¢S mmqo: :mmd. QZ< Bm¢m< :N mmBmHm :m mBHZ ZEDQOQ :m < ho m02¢2mommmm H manda - 8B _ 0:* : z 0.0 u- 0 -u 00000. 0.00 0.00 .. u- = 00* z z 0.0 uu o uu 00000. 0.00 0.00 u: uu : 00* : 0 @.H ll 0 II WQUOO. momm m0Q© II II 2 NJ _: : Hod II MNHHO. II 0 moOOH HWMDH II II : Hfii. : : II II MNHHO. II C m.OOH H. II I m :56 04* z 0 0.0 .. 00000. .. 0 0.000 0.M00 .. u :00 0 m : : NoH II mm NO. II 0 @330 .Jo HH II II 20.. nu : z 0.0 uu mmumo. uu o 0.00 0.000 uu uu 0020000 00m* = = -- u- 0 .. 00000. 0.00 0.00 -- u- = 00* = = 0.0 u- 0 -u 00000. 0.00 0.00 1- -H z 000* z o 0.0 u- 0 -u 00000. 0.00 0.00 .. u z 000* z = -- -- 00000. .. 0 0.000 0.000 .. -u : 0m* 2 z 0.0 .. 00000. .. 0 0.000 0.000 .. u- z 000 z 0 0.0 -- 00000. .. 0 0.00 0.000 -- u- = m* z = 0.0 u- 0 .. 00000. 0.00 0.00 u- u- a 0w* z z 0.0 -- 0 .. 00000. 0.00 0.00 .. .. msooawa 000* z a 0.0 .. 00000. .. 0 0.00 00.00 .. u- u- <00 = 0 0.0 -- 00000. -- 0 0.000 00.00 .. -- u- 00 z z -- -- 0 -- 00000. 0.00 00.00 .. -- u- 000 = 0 -- u- 0 -- 00000. 0.00 00.000 -- u- u- 00 z = 0.0 .. 00000. -- 0 0.00 00.00 -- -- -- <00 : o 0.0 .. 00:00. u- 0 0.00 00.000 -u -u u- :0 00000 .00 0 0.0 -- 0 .. 00000. 0.00 00.00 .. u- -- <00 make 9000 030 GH #50 CH mmmsm 0>dom madam 9:000 mpzcflz 003m commmmmflm oz 23m madam Dam 000nm h>mmm madam pnmflq pm dm\0mxpm so 009 mmnocH mmmnm pm sommmaoz QM ocoo mpSHom mpmm 300% oflhpmasao> mmdomo Hdpoe Aeoze0peoov 0 m0m DH: uu 0 .00039 0500300 no 00009 000 mmsad> DH: uu m .0000 0000 000 00003 :00. uu d z z 0.0 - 0 - 00000. 0.0: 0.00 . 00 0.0 -0. z 00* z 0 0.0 - 0 - 00000. 0.00 0.00 0: 0.0 -0. z 00* 0 = 0.0 - 00000. - 0 0.000 0.000 00 0.0 -0. z 00* z 0 0.0 - 00000. - 0 0.000 0.000 00 0.0 -0. z 00* z z 0.0 - 00000. - 0 0.00 0.000 00 0.0 -0. = 00* = 0 0.0 - 00000. - 0 0.00 0.000 00 0.0 -0. = 00* = z 0.0 - 00000. - 0 0.00 0.000 00 0.0 -0. = 00 = .0 0.0 - 00000. - 0 0.00 0.000 00 0.0 -0. = :0 z : 0.0 - 0 - 00000. 0.00 0.0: 00 0.0 -0. = 00 = 0 0.0 - 0 - 00000. 0.00 0.0: 00 0.0 = 0 z z 0.0 - 00000. - 0 0.00 0.000 00 0.0 = 00 z z 0.0 - 00:00. - 0 0.00 0.000 00 0.0 = 00 z = 0.0 - 00000. - 0 0.00 0.000 00 0.0 s 00 = z 0.0 - 00000. - 0 0.00 0.000 00 0.0 = 0: = : 0.0 - 00:00. - 0 0.00 0.000 000 0. z 0: 00000 .00 0 0.0 - 00000. - 0 0.00 0.000 000 0. 0000000 0: 0000 0000 0:0 00 0:0 00 00000 00000 00000 00000 00:00: 00000 000000000 02 sum 00000 000 00000 0>00: 00000 00000 00 00000\0m 00 000 000000 00000 pm SDNmmHoE Qa.dcoo mpsaom mpdm soak oHHpmSSHo> mmdobo H0008 t:l.| .00005000 0000-0000.0 00000 00 - * 000.0 - 00000.0 0000.0 0000.0 00000.0 00.0 00.0 00 0000.0 0000000 00* 000.0 - 00000.0 0000.0 0000.0 00000.0 00.00 00.00 00 0000.0 0000000 00* 000.0 - 00000.0 0000.0 0000.0 00000.0 00.00 00.0 00 0000.0 0000000 00* 000.0 - 00000.0 0000.0 0000.0 00000.0 00.0 00.00 00 0000.0 0000000 00* 000.0 - 00000.0 0000.0 0000.0 00000.0 00.0 00.0 00 0000.0 0000000 00* 000.0 - 00000.0 0000.0 0000.0 00000.0 00.0 00.00 00 0000.0 0000000 00 000.0 - 00000.0 0000.0 0000.0 00000.0 00.0 00.00 00 0000.0 0000000 00 000.0 - 00000.0 0000.0 0000.0 00000.0 00.0 00.0 00 0000.0 0000000 00 0000 0000 0:0 00 0:0 :0 00000 0>0mm mm000 00000 000002 00000 000000000 oz 0:0 000: 000 00000 0>00: 00000 00000 00 0m\00\00 :0 000 000000 00000 00 0000000: 00 0000 000000 0000 0000 0000000000 mm0000 00000 000000003 .0 .0 82 .0000 00> .0 .0 000.0300 .0 .3 200.0 0.040 .:mm ZOHBomm 053m mo Enough .mzoE gmomH gamma EBHB 0585» 20mm QHO< CHE-“5.4 UZHBOgvnm mMJ-dm :mw:o.o I :N 9.2 Baa :m u mmgmnEHm .3” mBHz 5.5400 :mNmJn .4 .mo 0.524205% HH mag-SH. ..._.~0...r\.0. — 10 @0000 0000000 Go 00009 mmSHO> :9: a: o 1IIIIIIIllllllllllilllllllltlln ©00N II II : II : : . Mm” H“. H” : ll : : ”H Momm m: . MUOWSMud HQ 00 m z n- z z 00 0 0000000 00 MH. H II II : ll : N.- _.. \ L : mflowflw. - mo.0 : 4m.0 n: : I: z 0 :m 00 cm 00 om. : 0 MM mwofl : FNoN ll : I... 2 z m. : m3. : 0 o 0 II II 2 : mm WH. EH Gm : MA. H : .: 2 Md.®.fi O o : OH 00.0 = 00.0 -- z -- = 00.0 00.00 w «M. = 00 Hotm : N270 .... : 1.. : : mm.wm . .0 : +04 MAM-H : aoofl ll : ll 2 : .JHoON : z : MH 00.0 = 00. -- . -- . 00.0 00.0 . z z 00 00. = 00.0 -- z -- z = 00.00 = z = 00 00. . 00.0 -- z -- . . 00.00 = . = 00 @004... : ma. ll : ll : mwom HP00 , NM. Om. : O SKUN : with II : ll : : mmowfl : : mw m®0NH : mmjrfi ll : II : Md.mH m®0m : : : ML Pm0m : ®H0N II : ll : : 1H.®N : W : W mF.H : mm. ll : ll : Hpom mwom : h : L 0m.0 : m0.m I: : an : : mw.m: : z : M 00.0 = 00.0 -- z -- = = 00.00 = = H 0 00.0 o 00.0 -- 0 -- 00000. 00.0 00.0 00 00. 0000000 0 0000 0000 000 00 000 00 00000 00000 00000 00000 000002 00000 000000000 oz 000 0000 000 00000 0>000 00000 00000 00 00\00\00 00 000 000000 00000 ph SONmmHOZ Q4 0000 005400 0000 300k 000000340> 004000 00008 gm .m .0 972. 048000 .2 .00 293 94.300 .mMBdS mBHz 4030044 4M2mHm xm4mOOD4 zma :BHZ 223400 :4 < mo moz<2mommmm HHH m4m¢a - 11 _ Griffith, Jasny, and Tupper (8), used a two-inch diameter sieve- plate column, 1&0 inches long, in a region of high-amplitude frequency range, separating cobalt salts from nickel salts for the Atomic Energy Commission. Their data were somewhat erratic, and the points which they plotted showed large deviations from the expected curves. This was probably due to the rather complicated multicomponent system with which they had to work. Belaga and Bigelow (9), reported data obtained from a sieve- plate column #5 inches long and 1.5 inches in diameter with one-inch spacings. For this work, acetic acid was extracted from the dispersed aqueous phase using methyl isobutyl ketone as the extractant. Graphs were presented showing the variation in HTUOE (Height of a transfer unit*) with frequency at constant pulse, and the variation in HTU with pulse amplitude at constant frequency. The data, although certainly indicative of trends, showed considerable variation within a family of curves. HTU values were found to range from 2.63 to 6.25 inches. §pray Columns Spray columns are another type of mixer-settler extractor which use only the energy imparted to the incoming streams and the density difference of the two phases for providing interfacial contact area. The dispersed phase is introduced into the column through one or more spray nozzles. This phase then travels through the length of column, remaining broken up into droplets. Essentially all of the mass transfer takes place near the spray nozzle, and increasing the length does not give appreciably greater mass transfer. Some spray towers are filled with packing such as Raschig rings or * Subscript "OE" refers to over-all transfer units based on the extract phase. 13 -12.. Berl saddles; these give more interfacial area, probably because of the extension of one of the phases into a film over the packing surfaces. ‘Although spray columns containing packing are more efficient than those without, the number of stages is increased only by about one theoretical plate for six or more feet of packing height. Pulsed Spray Columns An interesting report describing a pulsed spray column has been published by C. J. Billerbeck et al (l0). These investigators used a 1.5-inch column to extract acetic acid from water with methyl isobutyl ketone. The results of several of their runs are tabulated in Table IV. Packed Columns The packed column extractor has a complex mechanism of mass transfer. In these units mass transfer takes place not only by extended films on the packing surfaces but also through direct interfacial contact of droplets of the two phases. Efforts to increase the packing surface by using finer packing usually results in greater investment, because the capacity is reduced and the packing weight is more per cubic foot. Actually an optimum size occurs in any given column; packing smaller than the optimum is uneconomical because of high capital investment, and larger packing provides too little surface area for mass transfer. Packed columns are simple to build and easy to operate. The packing can be of almost any material and shape such as Raschig rings, Berl saddles, Lessing rings, spheres, Intalox saddles, spiral rings, or even gravel or cinders. About the only restriction is that the packing should be chemically inert to the liquids being contacted. It may further be pointed out that,as an approximation, the amount of flow varies inversely with the surface area of the packing. Besides the packing -13- .0093 0500300 00 0000.000 0000.000 PE 0 I... O mum. II. a II : .30on \LHoNQ COW II 0 mm. II : I... : Nméh mN0mw 03 : z 0 II C Hoofl ll .: ll : PMoNLm. moomw COM 7 : mw II 0 JM.H II : ll : PN.®M. mytoiw CON 8 : N. I... 0 WW. I... : I... : mP.~..m mm.mm OOm : : M ll 0 NOoH ll 2 I... : ONoPm \tmomo CON WFMMM. : L II 0 ©m.H II : II : 00.8 HP.:@ . O O : m -- 0 00.0 .: ._ -- _. 0.0.90 +0.00 0 0 .. 0 000.0 0 00.0 0000. 0000. 0000. 0000. 00.00 00.00 0 0 000M000 a 0000 0000 000 00 000 0H 00000 0>000 00000 00000 000002 00000 000000000 00 000 0000 000 00000 0>000 00000 00000 00 00x00\00 00 000 000000 00000 00 00\00Hoz 00 0000 000000 0000 2000 000000000> 000000 00000 .2030: .m..¢ 4000.0 mmmmmm .o .h. «fin—Hm .o .m «HHH £036,030 .6 «zommmdem .0... .0 20mm 0.8.0.4 £50,460 $4.340me $0.450 mBHB E04203 20mm 93¢ UHBmod OZHBOEXM 2....5400 Mgm 90de :m.H 0% 040 muggommmm >H .4433. - 1h - and a long vertical tube, only packing supports and distributors are needed for the construction of these columns. A great many studies have been made on packed columns to determine the effect of which phase wets the packing, how the choice of continuous phase affects the efficiency, etc. many of the packed columns reported have heights of a transfer stage from 5.0 to 20 feet. Smaller heights of a transfer stage are sometimes encountered in systems with good transfer characteristics, particularly when fine packing and small column diameters are used. Large packed columns are noted for channeling. Murch (ll), claims that the height of a contact stage is approximately proportional to the column diameter. Pulsed-Packed Columns Because of the inherent deficiencies of packed columns, it seemed like a natural step to go from packed columns to pulsed-packed columns. The packing could then act as an inmovable stirrer and the liquids, as they move up the column on the upstroke of the pulse, could smash against the packing and break up into fine droplets causing a large increase in interfacial area. This should in no way affect the continuous countercurrent flow of the two phases. Furthermore, the pulse could be supplied by a piston or bellows external to the column, for easy access. It is readily apparent that this up—and-down motion of the liquid should at least markedly decrease channeling, if not eliminate it altogether. Because the stroke of the piston must normally follow a sine wave, there are periods during the cycle when coalescence may take place. Coalescence followed by redistribution into drops with new surfaces plays an important part in efficient mass transfer. - 15 _ Among the first unclassified investigations made on pulsed- packed columns was that of Feick and Anderson (12), who used a l-l/E-inch column with 3/8-inch ceramic Raschig rings and l/2-inch McMahon saddles to investigate two different systems. The systems they studied were the extraction of benzoic acid from toluene with water, and the extraction of acetic acid from toluene with water. .According to the authors the efficiency of the columns was increased in some runs so the height of a theoretical contact was improved from nearly 12 feet down to about 7 inches. The authors attributed this improvement to an increased area of contact between the two phases under agitation rather than to an increase in mass transfer coefficient. They concluded this from experiments in which they replaced one solute whose major diffusional resistance lies in one phase with a solute whose resistance lies in the other; if the same improvement in extraction was found in both cases, it must be due to increased area. The results of several of these runs have been tabulated in Table V. Wiegandt and von Berg (5), published a summary of the work done on pulsed-packed columns up until 195%. In this article they reviewed the work done by two engineering students at Cornell University, P. C. Goundry and V. M. Romero. The authors tried to correlate data obtained by these men with that obtained by Feick and.Anderson. They were able to draw some general conclusions and to offer a comprehensive explanation of what actually occurs inside pulse columns. Their conclusions will be given later on in this report where they can be compared with the conclusions drawn by this investigator. u.—._,.l/\,.. .oflom vapoom Op domcmzo mpsaom :u . .mQCflh mflnommm :dwm. In “N .omdzm Oficdmpo why no woman man moddm> Dam nu o .mmdsm mzomswm mnp so momma mum moddm> DB: u: s -- omma. Nos. z moooo. omoo. os.w odmm com mama. z pmm -- owns. 3am. . assoc. owes. oa.w oa.mm co: mama. . saw -- ems“. Jam. = :aaoo. coco. om.w om.mm com Omsm. . sow -- ooao.a 0mm. z moaoo. osmo. Om.m ow.Hm o o = sad -- 03mm. same. : moooo. osmso. cm.@m om.sm co: = z and -- oasm. ammo. z :oooo. osmso. os.mm 0:.mm 0mm Omam. = spa -- osms. mmmo. s saooo. oamao. om.mm om.mm 0mm mama. = pea mam.a o :H.H ossmo. = wmooo. osmao. om.wm om.am o o z pma -- wamm. swsooo. z mwaoo. mmoao. 0m.ma mm.: 00m = = :H* mmw. swam. mmsooo. = Ammoo. mmoao. om.mm wa.a co: mama. a ma* mo.ma a os.ma mmaooo. z osmoo. mooao. cm.mm ms.a o o .. ma* -- ammo. wowooo. z ooaoo. mmoao. aw.» mm.w co: = = aa* -- scm . Homooo. z mmmoo. wwoao. o.ma w.ma 00m = = oa* -- smaw. Howooo. z wamoo. mwoao. w.ma :.aa com mama. z m* . -- a om.oa ammooo. z amoao. mwoao. m.:H H.:H o o = m* Mm -- a ma.m ommooo. = mmoao. mwoao. m.mm H.ma o o = s* . -- a mw.a ossooo. z cacao. Hmaao. m.mm m.~m oooa : z w* -- a mm. omwooo. = Lmoao. Hmaao. p.4m o.Hm oooa ammo. = m -- w ::.H wzsooo. z memos. :mmoo. m.H: m.mm 0mm z = s -- a as. ammooo. = oamoo. ammoo. m.am m.aw 000 z = m so. a mm. mawooo. . mawoo. ammoo. >.mm w.mm Om» mama. = m -- s cm.s ammooo. o.o mamoo. msmoo. «.ms m.aw o o mcmsaoa H pass owes pzo cH pso eH mmmem a>wom eased pemfin spans: smasm smmtmmmfla oz cam mama as: mmagm s>wom madam pgman pm am\hmmpm so Hum mmnocH mmsnm 9m zuwwoaoz pg omoo opzaom opmm 30am oflhpoezao> mmaoho Hmpoa zommmaz< .2 .m QZ<.Monh .0 20mm ¢B¢Q .Oom.wm mime .:mm.mm Hmchm 02H30¢m .mmed3 mBHz mZmDAOB 20mm QHO< UHONZMm UZHBOdexH mmADQ.sm mm.w o o z ma as.m .. mmaoo. amaao. msaoo. s aw.: ms.ma o o = Ha wo.m -- ssmoo. mmmao. smaao. e :m.om sm.om o o s OH mm.a -- msooo. mmaao. masoo. e mm.s :m.om = s s a mom. -- smooo. mmmao. somao. = so.ma Hm.ma z e e w ma.a -- «mace. mmaao. maomo. = ss.>m so.m e mm. s a wmm. -- Hsooo. oosao. wamao. e s s s Hm. s m mmP . ll moon—HOG. +~m HHO. mOmHO. : : = : NM . : m mo.a -- mowoo. smmHo. mamao. e a .. e an. e s mm.a -- smmoo. wmsao. Hmsao. e s z e as. = m 25 -- mice. 5:8. 3:8. ._ s .. __ o ._ m 0m.a -- mmmoo. mamao. memao. o ma.mm :m.om mma Ha. eassmso H poem seem ego sH sso sH swarm s>sea eased eeman opens: emasm eemseamsm oz cam mama Dam eased a>eem eased pawns pa am\sm\pa :0 sea mesocH madam pa so\meaoz pa eeoo eesaom spam seam esteessao> meaeso asses .emm smon: 02H20¢m .mmflbmom .2 .Q 20% 5RD .ESS mBHZ mmHmOd.HogHmBzomm mAmaB -19- . . mm. 2 OH . . ma.»a om ma a: m . . -- mmmso. waom. awed. mmmoo. om.ma aa.sa as omm e ma .JNLJ . PGOH. SMHH OMFOO . O O : PH 10m 0 I I gwio \ . FDHOO. OJ . SH NLH +1” WHQ. II NFCNO. mmmwflo UNOH. .V o Pmom mmow Nmm mmo. : WH. . ll WPmJO. MQHN. OmNH EOHOO . .. MN . xdfl @00. : mH. mam . m . ooaa. 0mmoo. mm m L m L . owm. .. wosmo :H a s . a . :m.m pad mmo. s :H MJMo II OPW+~00 Pmmnfio mmHHo mmmwm QMoW N300 mm NLHN. mnfl . -- mammo. Adam. baa. e . r, . . a WWW. I1 immmo. Pmmd. mPHH. OmNOO. mN.N mm.® mm Wmo. : NH «mo. .. moowo. omom. amaa. amaoo. om.m we a p Lem s Ha 00m. .. cameo. memm. cema. mwooo. ,m.m m.@ as maa. s OH wsm. -- mosmo. mmom. Peas. assoc. ss.w ms.m am and. s a saw. -- wmamo. anew. sass. mwooo. :e. ma.m mm mac. z m m s. -- osweo. osom. mass. space. ow.m sm.m a: smm. s a New. -- «memo. :mmH. mama. mmooo. mo.w mm.m a: ama. = o mos. n- macso. mama. adma. mmmoo. oo.w a .m a m . w d are m mam. -- omsmo. omom. moaa. amaoo. ma.w m:.w mm dam. s mmw. -- masmo. mama. emaa. mwooo. om.m as.m mm mmm. e a mam. -- wmsso. mama. cmHH. mmmoo. mo.w om.w mm ea . s m ems. -- oosoa. mmom. macs. mwooo. sn.c 39.9 4 m_H. e m e \ {s are aficmmao H usmwh 9.00%. #50 CH #30 CH m - g k3 .r - .2 - J . 7.. mean :8: madam h>soz madam pagan lmpwsmwwaxwwmzmgps.aq mpszwz Ir; .H OCOO @PSHOMW L L . wry. , .Hmvmw 00.: BQZ¢UHH3 . . one .osz co> . .sm.am amoHa: quxoam .mZoemz name .< .2.zo . . nwsbmomH . mm aged uZHaoamaxm mosz OHmomam zHaqumom :mm.o memmm%mmowmezmwwwz.zomm QHQ< 0Hamo< : mNm .- H. 5H0 WU 2¢zmommmm HH> mamas evapodnowhpopconnmo mm.m . mmm.a om mm.wm w Hozood< HassOmH O. o.m: Hw.o ON m mm mumomOHmm docwwmm o.m n- .1 aw.o -- .. eseomosem sesamem mmw o 0 ll mfifihogowhvnwp n: [monaso a ao> 0» mpdfimmga .. -- ampspflsp a Ho> om " :OH mGOsz HhPSQOmH we , «m os.mm osm.o ages : s.oa sacs o . o .o mess: I 0.4.. Om. wH HM. Om . . n. mums 0 1mm mama em .4 5 mm- o a a . . mmmaomfl no 3 ao\mm:an amps: new: apfl>esu 0. ages smmmwmmm apmmoowa>o ea Hg coamnoa HdwowwhousH owwfiommm mom%Hdm . . HES mugs .mcoapwmapmw>cfl on» ca vows who: heap sown: pm onfipmywmsop map pm oHQammom mm a anon H mm Umpmwfl whd hmflfi .HH> 6C6 «H> q> «>H a a A HHH HH H madame aw nzogm msmpwhw map ho moapammoym HmOHwkzm map mo pm -21.. Purpose It was the over-all purpose of this investigation to determine a basis for designing pulsed-packed columns from a minimum of experimental To do this, methods should be found to predict HETS and limiting data. capacities, possibly from experimental data on small laboratory columns. A number of experimental investigations were proposed either at the These include the start or during the course of the investigation. effect of the following factors on flooding velocity: 1. column diameter 2. packing density 3. flow ratio A. column height 5. pulse amplitude 6. pulse frequency 7. interface level direction and rate of mass transfer 8. They also include the effect on HETS of the factors listed below 1. column diameter 2. packing density 3. throughput rate A. flow ratio 5. pulse amplitude 6. pulse frequency 7. .liquid inlets Limitations and Scope This report confines itselt to those variables most important to the design and operation of a pulse column. It makes no attempt to go into many details which may -22- in turn be calculated from present-day knowledge. A list of the most important variables that would normally be encountered in a pulsed column investigation are given below: 1. 18. column diameter column height packing shape and size ratio of packing diameter to tower diameter material of construction of packing solvent-solute system flow ratio of the two phases end construction, and its accompanying end effects fraction of the total volumetric throughput pulse amplitude pulse frequency which phase, if either is continuous wetting characteristics of packing packing density packing support and fraction of free area direction of mass transfer concentration of solute in the two phases form of the pulse wave Using only 3 values of each variable and investigating all possible combinations would require 318 or 130,000,000 individual experiments. Since an average time of three hours is required for each experiment, this would require 390,000,000 hours, or at a normal working year of 2,080 hours, this is 187,600 years. Fortunately such a number of experiments is not necessary to be able to establish - 23 _ with a high degree of certainty how some of these variables change the column operation. 0n the other hand, it is not difficult to understand why this investigation makes no attempt to claim completeness. The experimental work here has been limited to: a. b. C. d. e. one type and diameter of packing one solvent-solute system one kind of tube construction one pulse wave form one type of packing support Furthermore, all of the possible combinations of the rest of the variables were not investigated but only those combinations which seemed most significant toward accomplishing the purpose of this work. All of these separate variables were investigated to some extent, but in a few cases only qualitative or limited quantitative data were obtained. APPARATUS AND PROCEDURE A schematic representation of the experimental pulse columns is given in Figure 1. This represents essentially the three different columns built, with the exception that the 3.32-inch diameter column has expanded chambers at both ends of the column so that an area is provided for the superficial velocity of the outgoing phase to decrease (see Figure 3). This decrease in velocity is reported as necessary by Blanding and Elgin (15), to prevent entrainment of the entering phase in the outgoing stream. Figure 2 is a photograph of the 2.127-inch column, and Figure 3 shows the 3.32-inch column. In this investigation, 8-mm lime-glass Raschig rings were used as Packing for all of the experiments. The characteristics of these are given in Appendix I. The 2.127-inch column held about 3 pounds of packing when filled to a height of 30 inches. The 3.32-inch column held about 25 pounds of packing when filled to a height of 105 inches. The exact height and weight of packing varied with the procedure used for settling the packing. The packing supports were l/h-inch high stainless steel rings cut from tubes with l/h-inch wall thickness. Strands of No. 16 B & S gage nichrome wire were silver soldered across the rings at intervals of 0.30 inches. This made a wire grid for the packing to rest on which had 50% free area. A packing "support" was also placed on top of the packing to keep it from moving up with the pulse stroke. This was held in place by four rods extending to the top of the column. A wire passed through the center of the packing and fastened to the bottom support. A.threaded bolt fastened to the end of the ' - .1. I. r":.'.' {1.8.11 C DEE“: I . ,_ . I‘i‘-‘l. . v-o-r . r JJA.‘ A Vent ‘gv Lupports ' 1’. 7,01‘ ll. , ‘I’lll'olt-‘l..clln|l‘.tl" r..l ‘31-!!! ‘l Illlc I- 1"}- “I, ‘!“i v 1 I I .. 1"- t‘ " -!.l ‘4‘.” I.‘ ' II... if .IIIIII . t Illl'ul‘lullulilull Sl.l|.lll'l|l!|ll I . txul..-l.|.l1 illl . . _ . _ x. \ 1» UV l\ illvull-l. . I ‘nl f A '- ON 14 k =1 Mlllll. \lillL ._ a; A 35“ Gal Organic feed -26.. Kim 2 . , .... v... u. a .. J _ . H. & - 27 - wire allowed the bottom support to be pulled upward in the column as the packing settled. his decreased the relative distance between the two supports and kept the packing from moving appreciably when the supports were fastened into place. The 3.32-inch column also had two packing supports constructed in the same way. These, however, had 61.7% free area. The expanded end sections on the large column were designed with a tapered Venturi-like approach to the packed section and tapered from 3.32 inches in diameter to 9.5 inches in a height of 12 inches. The water overflow was provided with a vent so no siphon action could occur to put a reduced pressure on the column. The water flowed from the top of the large column through 18~mm ID glass tubing to the drain. The small column used lh-mm ID glass tubing for the water exit. In each column the exit water line was made sufficiently large that little friction loss occurred to put back pressure on the column. The exit lines in both columns were designed for a superficial water velocity of less than one foot per second at maximum throughput. The CClu overflow in each column was provided with a vent to avoid siphoning action, and in addition was designed to permit variation of its height so that the interface level of the CClp phase could be positioned at any desirable place in the column. The adjust- able overflow leg discharged to a 250-gallon glass lined storage tank. The 3.32-inch column used l/2-inch ID Tygon tubing and connected at the column to l2—mm glass tubing. The small column used 3/8-inch ID Tygon tubing and 8-mm ID glass tubing connections. The adjustable leg had to be repositioned each time the CClk rate vfir In L) I fl'fi wk. 7“ s. n» H\- Vt. V)..- U. - 28 - was changed. In both columns the vents were large enough to handle the extra volume of liquid brought in by the upward pulse stroke, so it was not necessary to put in an expansion chamber at the top of the column. The CClu was pumped into the column by means of a stainless steel centrifugal pump* from a 250-gallon glass-lined tank. The CClu flowed through l/2-inch polyethylene tubing to filters that contained cheese cloth as a filtering media. From the filter, the CClu passed through l/2-inch stainless steel needle valves to rotameters. These rotameters were 3/h-inch size with three specially constructed floats to measure extremely small changes in volumetric throughputs. From the rotameters the CClu entered the top of the column through l2-mm ID glass tubing for the large column, and 8-mm ID glass tubing for the small column. In both columns the CClh was allowed to discharge directly onto the top packing support from the glass lines without using any type of special sparger or distributing weir. The water was pumped into the bottom of the column through the same size pumps, rotameters, and tubing. The water was stored in a 250-gallon stainless steel tank and a l/h-inch valve was used in the line leading to the column. Two D-ll stainless steel pumps were connected in series to both the CClu and water feed lines when large flow rates of these feeds were required. Two 250-gallon glass-lined tanks were used for the CClu. One tank was used as a collector for the CC1u discharged from the column, while the other was used as a feed tank. Both tanks contained built-in * Eastern Industries, Model D-ll. - 29 - agitators with stainless steel blades. They were also jacketed for constant temperature control. However, it was felt that the room stayed at a temperature close enough to 25°C that temperature regulation was not necessary. Tap water could not be fed to the column directly from the water main, because the water in the lines was under a pressure of 60 lb/sq.in. and contained a great deal of dissolved air. When the pressure was reduced, much of the air was released; this collected in the rotameters and made then inoperable. Therefore, the water was stored in a 250-gallon stainless steel tank where it was allowed to lose its dissolved air and also to come to room temperature. The pulsator used at the beginning of these experiments was a brass bellows of approximately 2-l/h-inch ID and contained six corrugations. The pulsation was produced by an adjustable motor- driven eccentric. The bellows was compressed on the upstroke of the eccentric causing liquid to flow into the column. On the downstroke of the eccentric the bellows would open due to the weight of the liquid resting on it. The bellows was sealed at the top by a No. ll rubber stopper. It was found that after several hours this rubber stopper would soften because it was in contact with CCIA. When the rubber was softened it would move up with the upward stroke of the eccentric and down with the downstroke. This would markedly change the amount of pulse volume which the eccentric was originally set for. It was therefore decided to replace this unit with a reciprocating piston in a h-inch ID nickel-plated cylinder. 9*- , yids" l .,..- ‘2.’ sfi. (7‘ (T‘ -r‘ ‘9“ ‘ll - 30 - The leather seal used for the new pulsator was a standard replace- ment part for a reciprocating water pump. Inside the cylinder and below the piston a small drain was installed through which the small amount of CClu that leaked past the leather plunger was drained. The outlet at the top of the cylinder, leading into the large column, was 5/8-inch ID x 3/h-inch OD stainless steel tubing. The entire inner surface of the cylinder, where CClh came in contact with the metal, was nickel plated to prevent corrosion. The amplitude on this type of pulsator was very reproducable and the measurements which were made at the beginning and end of each run were always found to be the same. The stainless steel tube leading out of the top of the pulsator was connected to another stainless steel tube of exactly the same size. The latter went through the rubber stopper into the bottom of the column and projected two inches into the column. The two stainless steel nipples were fastened together with a short section of B/h-inch ID polyethylene tubing. The rubber stopper in the bottom of the column was covered with a l/l6-inch layer of mercury to prevent corrosion by the CClu. The entrance from the pulsator into the small column was glass tubing l2-mm ID x lS-mm OD x 5-inches length. The form of the pulse wave supplied by the piston was essentially a sine wave, since the eccentric covered the 360° cycle in a uniform manner. A three-speed pulley was attached to the shaft of a l/h-HP motor that had a speed of 1750 RPM. The 3-speed pulley on the motor shaft was in turn fastened to another 3-speed pulley by a V-belt. The second pulley drove the shaft of a speed reducer that had a speed reduction ratio of lO.h. This allowed the speed of the eccentric (7‘ - 31 - to be varied from 65 to 215 RPM. .All of the materials of construction for these columns had to be corrosion resistant because CClu saturated with water is extremely active and attacks the less resistant metals. Iron, steel, and galvanized surfaces are quite poor in this respect. Polyethylene becomes brittle after contact with CClu for three to six months. Tygon appears to be slightly more resistant to CClu but becomes brittle after approximately six-month exposure. Solvent-Solute System The system carbon tetrachloride - water - acetone was selected primarily because at solute concentrations up to one percent the distribution coefficient is essentially constant. The components are also easily obtained and offer little fire hazard. The experiments were always started with a mixture of carbon tetrachloride* containing approximately 1% by weight acetone**. The acetone was extracted from the carbon tetrachloride with tap water that had been allowed to lose its dissolved air and come to room temperature. The extract, after sampling to determine its acetone concentration, was discarded to the sewer. The raffinate, which was saturated with water, was sampled and discharged to a 250-gallon storage tank. When all of the feed solution had been used, the agitator in the storage tank was turned on for 30 minutes. At the end of this time the carbon tetrachloride solution was analyzed for acetone content and enough more acetone was added to bring the concentration back to approximately 1% for another run. Analytical Procedures Several methods were tried for the quantitative analysis of acetone in water and in CClu. These methods all used the same basic step of titrating the HCl released when k * Dow technical grade. v - 32 - acetone reacts with hydroxylamine hydrochloride. CH CH 3‘c = 0 + HONHé'HCl 3‘s = NOH + HCl + H20 I I The method was first proposed by Hoepner (16), and later investigated more thoroughly by Marasco (17), who found that certain conditions had to be carefully controlled in order to obtain good results. Bennett and Donovan (18), did further work on this analysis. Bryant and Smith (19), proposed a method of analysis using a pyridine solution with bromophenol blue indicator to determine the HCL liberated. All of these methods were tried by this author and the one selected was essentially the one proposed by Bennett and Donovan. In this method, 12 g. of hydroxylamine hydrochloride are added to 6000 ml of tap water. Sufficient methyl orange indicator (about 5 ml of saturated water solution) is added to give it a golden yellow color. Either acid or base is added to this solution to bring it to the neutral point. For the tap water at Michigan State University, 31 ml of 1.0 N HCl is required. Approximately 600 m1 of this solution is then poured into each of two lOOO—ml beakers. One of these beakers is considered the standard color, and to the other is added 20 ml of the solution to be analyzed. The HCl liberated by the acetone is titrated with 0.2 N alkali. Either NaOH or KOH can be used. The amount of alkali needed to bring the color back to that of the standard solution in the other beaker measures the amount of HCl liberated. This, of course, is directly proportional to the acetone present in the 20 ml sample. Since the reaction only takes place in the water phase, it is necessary to supply vigorous agitation to the solution when CClu is -33.. being analyzed. Even in analyzing the water phase, this method takes several minutes because, even though the solution is initially titrated to the neutral point, the release of HCl is slow and more is usually given off after a few minutes standing. The time necessary for this analysis depends on the quantity of acetone originally present; about 10 minutes is required for the average determination. A.refinement of this procedure was brought about by the intro- duction of an electrometric titration apparatus. In this technique, the pH of the standard liquid is determined by a potentiometer. The liquid is constantly stirred by an electric stirrer provided with the apparatus. The standard solution is then removed and replaced with the solution to be analyzed. Alkali is added slowly and with constant agitation while the pH instrument indicates the acidity. Enough alkali is added to bring the pH back to that originally determined for the standard solution. One note of caution should be mentioned when using this procedure. The 600 ml of hydroxylamine hydrochloride is intended to provide at least 50% in excess of the amount required for the acetone. However, if more concentrated acetone solutions are used or larger samples taken, then correspondingly larger amounts of the amine hydrochloride solution should to used. The analytical procedure was tested over the entire range for which it was expected to be used in this investigation. These results were reproducible and checks for known acetone concentrations were satisfactory. The following is typical of these checks: Four lO-ml samples of 0.17h-normal aqueous solution of acetone were analyzed; two in - 3h - the presence of 20 ml of carbon tetrachloride and two in water. The samples were placed individually on the electrometric titration apparatus. The samples in carbon tetrachloride used 38.5 ml and 38.8 ml of 0.0h5h N sodium hydroxide, while those in water used 38.h and 38.6 ml. These values gave a maximum error of 1.15% for the 1.7% millimols present in the original lO-ml samples. This agreement was considered to be excellent for the proposed investigation. Operating Procedure To begin a series of extraction runs, the column was first filled with water, then Raschig rings were poured into the top of the column and allowed to fall down through the water and settle on the bottom packing support. When the rings had filled the column to the predetermined height, the top packing support was put in place and the top rubber stopper was fastened in to prevent the upper packing support from moving. Only unpulsed runs could be made using this procedure. If it were desired to make pulsed runs, the upper packing support was left off and the water in the column was replaced with CClh. Then the pulse was turned on and allowed to run for a period of time depending on the degree of settling or packing density desired. The reason for filling the column with CClu for settling the packing can easily be seen if one recalls that CClu has a density of 1.59 g/ml and lime glass has a density of 2.2 g/ml. Because the density difference is much less between glass and CClh than it is between glass and water, the rings are more mobile and free to rise and fall with the pulse stroke when immersed in CClu. When the packing had reached the desired density, which could be determined by weighing the amount -35... of rings that were added and knowing the volume they occupy, the top packing support could be put in place and fastened with the rubber stopper. A.long wire, fastened to the bottom packing support and running up through the column to a stationary plate above, was tightened by turning a nut on a threaded bolt fastened to the end of the wire. This had the effect of raising the bottom packing support and clamping the packing in a non-movable position. A typical settling cycle for the 3.32-inch column involved the pulsator being set at lO-mm amplitude and 65 RPM and allowed to run for 12 hours. .At the end of this time the frequency was turned up to 125 RPM and the pulsator was allowed to run for an additional three hours. The packing density changed from hh.5 to h9.8 lb/cu ft during this period. The pulse amplitude referred to in this report is the total millimeters of travel, or the sum of the up and down stroke measured in an empty cross-section of the column. The piston pulsator was capable of 50 mm of amplitude, but it was never set for more than 10.5 mm because it was feared that the glass column was not of sufficient strength to withstand this much pulsed volume. In fact, four end sections were broken during this investigation. The CClh, which contained approximately 1 weight percent acetone, was pumped from the 250-gallon mixing tank through the rotameters and into the top of the column. The CClu was previously saturated with water so there was no gain in volume of the CClu flowing through the column. On the other hand, the water was not saturated with CClh in the feed tank, but the curvature caused in the operating line due to the increase in volume of the water phase was insignificant. _ 36 - It should be pointed out that the rotameters, although calibrated for throughput, were never used for measuring the flow through the column but were only used to indicate any change in the volume of flow that occurred during a run. Rates were measured by collecting the two outgoing phases in 5-gallon bottles for an interval of time that was measured with a stop watch. The two phases, after weighing, were then blended thoroughly and small samples were taken from each for analysis. In most of the runs the column was operated at its maximum throughput capacity. To reach this maximum capacity the flow rate of one of the streams was raised in small increments until an inter- face appeared above the top packing support, below the bottom packing support, or in both regions. Such interfaces occur because all of the stream entering the column at that end cannot flow through the packing under the conditions of operation. If the adjustable CClu overflow leg is kept in a high position, an interface will ordinarily appear only at the top, and if it is kept in a low position the inter- face will ordinarily appear only at the bottom. In the runs reported in this thesis, the overflow leg was adjusted to an intermediate point such that interfaces occurred at both ends simultaneously. Capacities obtained in this way were greater than those which would have been obtained if either the top or the bottom interface alone was allowed to limit the operation. The column was always run at such balanced flooding conditions except in a very few tests where the rate was decreased intentionally to see what effect this might have on HETS. The flooding runs are identified by the letter F and those below flooding by R. -37.. The column was started by first turning on the pumps leading from the water and CClu feed tanks. The needle valves in the two lines were used to adjust the volume of the flow of each phase through the apparatus. It was usually easier to set the CClu rate at some given throughput on the rotameter in that line. The water rate was then gradually increased until the appearance of an interface at the top or bottom of the column. When this interface appeared, the height of the adjustable CClu overflow leg was either raised or lowered until the interface disappeared and the water rate increased again. When interfaces appeared at both top and bottom the leg was readjusted; the lower interface was positioned by simultaneously increasing or decreasing the water rate. A time of 30 minutes to one hour was usually required for adjusting the interfaces to their necessary positions. One of our requirements of column operation was that the inter- faces should all remain immovable for at least one hour of constant operation before the readings and samples were taken. That is, if one of the interfaces started to move into or away from the packing, making it necessary to change the rate of flow of one of the phases, then the time had to be restarted. This was done to be sure that all of the contents of the column were in a steady state condition. Because of the expanded end sections on the 3.32 inch column, more time was required for it to reach steady state than was required for the small column. If the run was to be pulsed, the pulsator, which had been previously adjusted for amplitude and frequency, was turned on after the appear- ance of the two interfaces. Because pulsing usually changed the throughput rates, it was almost always necessary to make more adjust- - 38 - ments in the rates of the two phases as well as changes in the height of the CClu overflow leg. The pulsed column was again brought into balance by the method mentioned previously and operated for one hour before recording rates and taxing samples. The exit water and CClu lines were each equipped with two l/2-inch brass cocks. When the column had run the necessary time, the valve in the water line leading to the drain was closed and the valve leading to the 5-gallon weighing bottle was opened. The time to fill the bottle was measured by stopwatch, and after the quantity in the bottles was weighed, the rate was calculated. At the same time that the water sample was taken, a sample was also taken of the CClu. A 250-ml sample of each of these two phases was placed in a glass stoppered bottle to prevent the acetone from evaporating. At the beginning of the sampling procedure, readings were taken on each of the rotameters, the height of the CClg outlet leg was measured, and the height of the interface in the 6-mm glass level gage on the side of the column was read. The small samples of each of the two phases were analyzed and the amount of acetone present recorded in millimoles/liter. Knowing the flow rates of the two phases and the amount of acetone in the incoming and outgoing streams made it possible to make a material balance around the column. If this material balance did not check within 5%, the results were discarded and the run repeated. Accuracy and Reproducibility ,An attempt was made to determine what kind of accuracy could be expected from these experiments. This would, of course, be reflected in the answers obtained from a series of experiments which might be expected to give the same rn -39- answer. As an example, the HETS (height equivalent to a theoretical stage) appeared to be independent of the flow ratio when the column was operated at flooding in the manner of operation described in the preceding pages. When the HETS values found for a series of 10 runs on the small column, which were identical except for the flow ratios, were compared, they were found to have a maximum deviation from the mean of 16.21% and an average deviation of 8.11%. If the arithmetic mean of 58.72 inches can be considered as the correct answer, then the standard deviation is 5.57 inches. A.similar comparison was made of the HETS values obtained on the 3.32—inch column for 10 identical runs with varying flow ratios. Here the maximum deviation was found to be 20% and the average deviation lh.0“. If the arithmetic mean of 62.97 inches is the correct answer, the standard deviation is 8.25 inches. The results quoted above are the most erratic results obtained, because they were made on unoriented packing. The packing appears to have a tendency to orient during the first series of experiments that are made after the packing density has been changed. This phenomenon will be discussed later under Experimental Results. It will be sufficient here to point out that after these initial trials had been made and the packing had become oriented, the results stabilized and were much more reproducible. Illustrative of this last point is a series of five trials made on the 3.32-inch column after the packing had become oriented. On these five experiments the maximum error was 0.138% while the average was 0.076%. The standard deviation calculated in the same manner as the others is 0.h82 inches when the mean is 57.9 inches. - ho - One of the reasons for the choice of the system water-acetone- carbon tetrachloride for use in these experiments is the fact that the equilibrium line is essentially straight in the dilute region. The line does, however, have some slight curvature. Assuming a straight line makes little if any difference where only a few theoretical plates are obtained, because in this case the operating and equilibrium lines are far apart. With a large number of plates, the operating and equilibrium lines are close together. If they are straight and parallel they do not have to be as close at any one point as they would if consider- able difference in shape or curvature existed. Small errors in positioning of the operating line, therefore, lead to much greater errors in the HETS when the NTS in the column is large, and this effect becomes even more important when the operating and equilibrium lines are not straight and parallel. These factors were recognized and taken into account before beginning experimental work. The columns under pulse turned out to be much more efficient than had been originally anticipated, thereby giving a great many stages in the column. This necessitated making the operating line and equilibrium line essentially parallel. Actually, conditions were chosen to make the operating line somewhat closer to the equilibrium line at the dilute end, where small per- centage errors in analysis would not have as much effect on the distance between the two lines. This was also convenient, because operating in this manner made it not so important to correct for curvature at the concentrated end. - hl - Several runs were selected where the HQO/CClh flow ratio multiplied by the distribution coefficient had a value of 1.0 to 1.1. These values are tabulated below in Table IX. TABLE IX Dependence of Stages on Exit Raffinate Concentrations Run No Flow Ratio Final Conc Stages Fihh 1.1 121.0 1.37 F1u6 1.1 28.3 7.30 F152 1.0a 21.7 9.9 F152P 1.05 18.2 11.35 F153P 1.007 15.0 17.6 F152P 1.005 16.5 16.35 F137 1.02 113.0 1.77 Packing height, 101 inches, in 3.32—inch column Packing density, 50.h lb/cu ft CClu/H20 Flow Ratio, appoximately 2.1 When these values are presented graphically in Figure h, it becomes evident that, when many stages are present, a very slight error in analysis of the raffinate stream would be greatly magnified in the number of plates. Method of Calculation The data representing the efficiency of column operation can be expressed in either transfer units or theoretical stages. The latter has been used for the calculations in this thesis; however, many of the experimental results have been calculated using both concepts. The arguments for and against this choice have been left for presentation under the heading, Discussion of Results. 2......15 ,.._ :22... .. ,1 . .. 1 L. i . 1 . .. - ._. .1- _,. l . 1. . .1. $511 - -+: l M 1 17.14 m .1... 31111131111315.11511113. ’ v 1 llAl I .Ilvl‘ll . 111’. v 1 51.... 11.-..-- .4 - - -.. 5—,... __._. M-.- , . --- HLw—xua 1-.. r . .~ .~ “1.411%..1. ‘ 3 i1 - a ’ .7777 1 . 1 1 j.. ., 1.1.... - ..1.. .1. 1 H“vw «7* w“ » . . . . ... W1.H__+..__._w - . . . . 1 1 ..... .1IT' ' . I .i... . A 7‘ 4 . . , 5....-.a...‘ >— - § . . . . ..: , . . . . . . .- 1 -, 1 . . , a 3 . . “Hp—.VH—O—H Wr-H—a . . . . . . . . . . . . 21.; V -112- 1 1 1 . 1 ———o— L 4 ---—1r..... -q..r-‘i._.. i... . . . ‘ . r1 7|. fi-‘l‘lhlg . Q . . . . .1;-—~—1-—_.-45__..-..0.'.-- ——1T . -v—_..1i—.. ...__ b.—.—. 4r.— 1 . . ~--- 41...; . . --.-._.L.l . D 1 1 ! p 1 . 1 . . . a A]..- . . --.. 1...- 1.... .1 m; 121111 1.191119 . v A ' ‘. 3'2}: ”.851 11?;T1mis1 .91 7.3 4 O - h3 - The theoretical stage concept considers the column as consisting of a number of equilibrium contacts or theoretical plates, called the number of theoretical stages and commonly abbreviated NTS. When the NTS is divided into the height of the column, the height equivalent to a theoretical stage, or HETS is obtained. In all of the runs made in this investigation where HETS values were calculated, the inlet solvent (water) had a zero concentration of the solute (acetone) when it entered the column. In the calculations when only a few plates were obtained, the equilibrium line was considered to be straight and to have a slope equivalent to 2.135 parts of acetone in water to one part of acetone in carbon tetrachloride. In runs of low NTS, the number of theoretical stages may be calculated analytically using a method originally derived by Kremser (20) for gas absorption. This method was reviewed and expanded later by Sounder and Brown (21). In the derivation, two hypothetical plates, in addition to those in the column, are considered to exist; one above the column and the other below. The top hypothetical plate has the gas leaving the column in equilibrium with the entering liquid, while the bottom plate has the leaving liquid in equilibrium with the incoming gas. A.material balance is first made around the top of the column, followed by an equilibrium step. This stepwise procedure is continued going down the column until a general expression is Obtained for any plate. The general term is then combined with a material balance around the whole column, and when the assumption is made that the entering absorbent contains none of the solute, the resulting equation can be simplified and rearranged to give: 4 - _ - ao- ‘ . n- ,. 1--. .1» v-' a ~ —. -0 - L .- m.‘ -- 1 r . . . ....-.I .' . F‘ n ..-..-v.. u o.. .‘l ... av Du... ‘- U. )1. a . u». - a . . 7“ C‘ .- h""". v. C . - o 1‘ "'3 “‘ ‘- ~€ "u.“ r1 ‘\' .‘~--.'. . I - . -1131... (111:3 + 1) 1n m/R = ln [1 + (m/R — 1) Yl/Y J / 2 In this equation, NTS represents the number of theoretical stages; Y1 and Y2 are the concentrations of the solute in the solution being extracted (carbon tetrachloride) at the inlet and outlet respectively. The extraction factor m/R, is the ratio of the lepe of the equilibrium line to the slope of the operating line. The slope of the equilibrium line, m, is the distribution coefficient and the slope of the operating line, R, is the ratio of the flow of carbon tetrachloride to the flow of water. When the equilibrium line cannot be considered straight, the number of plates may be counted on a McCabe-Thiele diagram as shown in Figure 5. The only inaccuracy in this method is that a proper numerical value for fractional plates cannot be obtained. 11‘! -115- EXPERIMENTAL RESULTS Distribution of Acetone in Water and Carbon Tctrachloride In order to obtain distribution data corresponding as nearly as possible that occurring in the column, equilibrium determinations were made on carbon tetrachloride samples containing acetone which were taken from the CClu feed tank. Measured volumes of these were added to glass-stoppered bottles containing water and carbon tetrachloride in known proportions. A feW'Raschig rings were added to each bottle and then the bottles were sealed and placed on electric shakers. After shaking for two hours, the bottles were removed and a potentiometric titration was made for the acetone content in each phase. Knowing the volumes of both phases and the amount of acetone added, it was possible to obtain a material balance as a check of the analysis. These data are given in Table X. The tests were all run at room temperature. TABLE X Distribution of Acetone in Water and Carbon Tetrachloride Distribution Acetone in Water Acetone in CClu Coefficient 85.5 38.2 2.211 209.0 96.6 2.16 380.0 181.7 2.09 529.0 260.5 2.03 680.0 3110.0 2.00 The acetone concentration is expressed as millimoles/liter of solution. The experimental points are shown on Figure 5, as well as the smoothed -Lq- curve through the points. Coordinates for the smoothed curve were analyzed by Newton's method of differences and are shown in Table XI. TABLE XI Distribution of Acetone in Water and Carbon Tetrachloride Acetone in Acetone in AK AX Distribution water cc1h l 2 Coefficient 0 0 -- -- 0 100 15 - A5 -- 2.222 200 92 - A7 2 2.171 300 1M1 - M9 2 2.128 #00 192 - 51 2 2.083 500 215 - 53 2 2.011 600 297.5 - 55 2 2.017 650 326.0 -- -- 1.995 The acetone concentration is expressed in millimoles/liter of solution. Seidell (22), reports a value of m equal to 2.233 at 186 millimoles per liter of acetone in water and 2.205 at 322 millimoles per liter of acetone in water, determined by Herz and Rathmann in 1913. The corresponding values from Figure 5 of this thesis are 2.228 and 2.133. As a further check of the distribution coefficient, several values were calculated from runs which were known to pinch at the concentrated end of the column, assuming equilibrium existed at this end. These values checked very closely with the ones obtained in Table X. The average value of m in the range of concentration used in this investigation is 2.135. This is the value used for all calculations where the distribution curve could be considered a straight line without noticeably affecting the accuracy. - 88 - Small Column Operation Runs on the 2.127-inch diameter column may be divided into 0three types: unpulsed runs on loosely settled packing, unpulsed runs on settled packing, and pulsed runs on settled packing. Unpulsed Runs on Loosely Settled Packing Runs F56 through F63A appearing in Table XII, were unpulsed runs on loosely settled packing. These runs were made at flooding to determine the through- put rate and HETS as a function of flow ratio. It can be seen in Table XII that the sum of the square roots of the velocities of the two phases gradually decreased after the first few runs. This was attributed to contamination collecting on the surfaces of the packing, in the form of a dark brown stain. When the packing was washed with concentrated hydrochloric acid, the color disappeared and the rates measured after this were found to be in accord with the original values. For the rest of the investigation the packing was acid treated after approximately every two days of continuous operation. The remaining runs in Table XII are proceeded by an R, indicating that they were not made at flooding but at reduced throughput. These runs, utilizing only part of the column capacity, were made to determine what effect reduced throughputs might have on HETS. sod and Unpulsed Runs on Settled Packing Table XIII lists both pulsed and unpulsed runs on settled packing for the 2.127—inch column. To settle the packing, the amplitude of the pulsator was set at 8 millimeters and the frequency at 125 RPM. The top packing support was removed and the carbon tetrachloride in the column was \/ .oopnmmoh mm: nth map a emanawnoonoo npflz dfizaoo esp mnfinusam Mmpmd .mfih mflap do hphfio ma: unfit am map ) .C I... .x. .unfixodm exp mo mdwowa map #6 dam: modanopnfl Spa: mafia unadump l: m .wnfldooa% pm can nu m or\. \flaWo II o o _ . m.oL Urn. m 0 0mm 0 o; am can :ooa awe : m: wmm -- m.pw o.omm o.m» ma. mam wma mam o . o In- t \ 0‘) o 0.:w maw. aa 0 mum 0 mm ea new owed age 0 mm mpg -- m» o.mmm 0.0m mm. mam ,mw gem o.mm mph. n. ma o.mmm o.mma >w. 3mm 0mm mom n.0m mmw. -- m. 0.00m o.mma Ho.a mam Ham 3mm :.:m awn. m.aw om o.wum 0.:m ma. 0mm amma «mom m.m: owe. m.om we o.mem 0.0: am. 4mm oaaa mma* m.mw meg. 4.am moa 0.0mm o.mm me. me mama mwm o.mm adm. m.am Hma o.mam o.maa om.a mead mew ance o.mm awn. m.mm OJH o.m:m o.ama oo.a mmoa owe Hma* m.:o mag. o.aw m:a 9.0mm o.ama oa.a m>m oww owe o.mm mam. m.aw mma >.mwa 0.:em m:.> mmom mam mma m.am 0mm. m.ao m.®sa a.mw o.mom Om.m oewa aim mam m.om Omx. m.Hw mmH m.wwa m.®:H @H.m :mmd mm; yum e.Hm “me. a. e wma >.mma m.wHH oa.m owma Omw one mmzoeH mam mmpam 950 l aH pso am\aaoo :Hoo nope: on cam mama 30am anoom l none: capam 20am sasxaa .wm to 53m nopmwxn x 2 ocoo mpzavm noadm loam afimhossao> an :o\pa H.e: weHmzam UZonem QHmADmZD HHN m4m<8 .zn.om Hmonm ozHaoam .oZonam amqseamza :eHx zazaoo :mmoo.m was no mozezmommmm .HDMH Swim»... fiOSdeU mama... QESHOO was. CHRH mHJP .HOQth ll * w.mm mm». o.am wmm mm mma m oma oema imam >.m: mm. o.mm mam em 0 o ova 0mwa mwm -- -- -- -- - o o mom HOQH . amwa m.0m Odo. m.mm mam me o 0 New nmsa maa* N.®© Had. O.m® O.an 0w 0 O 03% Haw fiwm m.»a ma.a o.mm o.mma m.wma mma m ooaa mww ,maa m.am sax. «.mo o.mma mo 0 0 am» owe man m.om Om.a m.m> m.mma 0.0mm mma m :mma 02m mmaa o.m» Hz. maa m.mma P.0aa o o omma mma mam m.mm om.a m.maa m.Mma m.3>m mma m 3mm mmm ease m.om em. m.ama m.mma - o 0 0mm mm Hwn* ~.mm mmo.H mma m.mmH mom nma m cenm 0mm moan >.m: mhw. cam wwm man 0 o omma 0mm ova mozosH mzz 950 + nH pso opsmaz ea :Hou naan on cam mHm: lauo Hugs: How omadm cfla\aa nogflq\ooaoz E 0:00 opfiflom mmdomo Hmpou moudm kodb ofinwmazdo> moeemqsm meg mom ammo macaque menace ne.0m emoHam UZonem an so\pa a.am we mean oszoam UZHmoem amezmHmoz:-Qmaeemm 29H: zzpqoo =ema.m was go meme oszooaa mmmameD mz< memqsm HHHX mAMmemsmmn :H amen: one + oQOpmom eH + H00 omN mom amps: map + mdoemom RH + :Hoo madam pcmmmmh mH OmN mmH Hops: amp + ocopmom RH + :Hoo deHHpde NH OHm 00m Hops: mew + snowmen &H + :Hoo Hdemmp HH omm OmH Hops: map + mammmopm BORN hmppzh sH mssom NH comem mmopoom flH + :Hoo macaw pcvmmoh OH an CNN amps; map + momma QH mason NH deacon mnopoom fiH + :Hoo odmhm pnmwsmn m omw ON: nope: any + onlompohHom cH mason NH dodem mdopoom &H + :Hoo madam pnmmcon w u: owwH Hops: gap + snowmen &H + :Hoo Hmstmn e :NH om amps: amp + odopmom flH + :Hoo 3o: w owH ONH Hops: map + ocopood RH + :Hoo mddnm pcowsmh m QOHdeeo on Hops: as» + :Hoo madam pcommmn : cOHmHSSo on pope: emHHHpmHe + :Hoo memnm pcomson m OQOH 00m mops: eoHHHpde + :Hoo andwon .N OQOH coo Hopes map + :Hoo* pdemoh H H8 OH H8 m mHmHhmpmz oz mzm Sodom 0p accoomm CH oEHB ZOHBHX WHmHx mqmee I u... YE..." "‘i’v1.‘..ln .I -55 _ a series of runs were made where acetone was left out of the system. These runs appearing in Table XV were made on well settled packing on the 2.127-inch column at various flow ratios. In part of the runs, a small amount of acetone was present in the carbon tetrachloride. Those runs made after F80 contained no acetone in either phase. Pulsed and Unpulsed Runs Containing Acetone ' Once again acetone was added to the carbon tetrachloride feed and runs were made on the 2.127-inch column at a packing density of 51.7 lb/cu ft. These runs, appearing in Table XVI, show a decided increase in throughput rates, apparently due to the presence of acetone or to its transfer from one phase to the other. In Table XVI are also runs showing the effect of frequency on throughput rate and HETS at constant amplitude. Appearing on the same table are results which show the effect of amplitude on throughput rate and on HETS at constant frequency. Pulsed Runs Utilizing Only Part of the Column Capacity Previously, runs were made showing the effect of reduced throughput when the column was unpulsed. Table XVII lists trials made showing the effect of reduced throughputs when the column was pulsed. For these runs, the carbon tetrachloride-to-water flow ratio was held essentially constant at 2.1. It will be recalled that the amount of possible throughput was greatly decreased when acetone was not present in the organic phase. In order to determine whether this effect was due to the presence of acetone or to its rate of transfer, several runs were made in which two times as much acetone was present in the water phase as in the carbon tetrachloride. In other words, as much acetone was present in TABLE XV RESED.AND‘UNPULSED FLOODIHG RUNS ON TE PACKING HLIGLT 30.7‘" PACKING DENSITY El. 7 lb/cu ft ACETOZTE III TED?) ORGANIC FIJI ‘D 2.12?" COLUMN} JITH SH FILED PfiCKING mnmgmmmmsor Volumetric Flcw Rates ml/min Flow Ratio Sum of Run No Water 0011, comp/Ego Sq Roots F77 81:.0 11.06 o . M“ 3 2:9 .11; ”(T 1.06 1.640 1 .08 3 L21 . 13+ F78 366 9 51+ 2 . 60 :0 .02 F782 102 5 ‘32 3 . 08 37' . F79P 1360 128 0.093 93.15 FTTA 780 9123 0 . 972 ’49 . 02 PITAP 373 um 1 .19 1.0.353 Fran. 560 862 1. 97 ~ 93.36 F79A 911.9 133 0.191 132.28 F79AP 1290 1136 0.133 1:63.23 +F6‘0 690 we 1.1% 92.82 F80 - 1 650 324.3 0.928 1+3;..02 POOP - 1 67; 288 0.127 ~+3.00 FBOA 670 173 0.298 39.09 FBOAP 672 282 0 . 920 J 2 . "(0 F803 68; 193 0.282 1.008 F800 610 1233 0 . 760 ‘45 . 22 FGOCP .330 310 O . M92 3.2 . TO FBOE 660 190 o. 227 .99 FSOEP 550 300 0.1232 :2. d2 FBIP 969 213 0.216 216.00 F82 278 7650 2 . do ti; . 60 F822 271 160 l - =36 37 . 6'? FSOF (3‘40 193 0.22" 37.29 FGOFP 636 307 0.1.62 3.2.73 F800 61.0 193 o . 302 39 . 20 FSOGP 636 308 0.1.09 M236 * -- after this run, the carbon tetrachloride feed contained no acetone. P -- pulsed at 69 RPM and 9 MM amplitude .dfisaoo map CH pmflom gosflm d donodmp mflnp :- ** .mmfidooam zoaop memuflam n: * rum T mwh o.mm Hw.o o.mmm 0.9»m mm; o o 2m; mmm mddm o.m m .m 0.01 m.m:m :wm mam m mrma oao mflmm m.oa om.a o.oaa m.m2m om: mma m omca 04m mumm H.mm QH.H o.wma m.mIm mam mo m omma 09m mum 2.2m No.0 0.022 m.m2m mam o o omwa 0mm m+m2 -- -- -- 2.02m -- no m OLF no mum -- -- -- m.®rm -- mm. m 022. OHM Iuw. -- -- -- «.mmm -- o o omma o.m mommm I .. a on H {I II II II m.®:m II mmd m 0rd at muHum . . m H m Oma OOIH L t. u- I: n: m new nu mm m Oma oww madam II II II mowfio II L rN MIA... Odmfl fimwrm -- u- I- m.@2m .. o o mwa ogma mqaom -- -- -- m.w2m -- mm m yma coda moawe II II II A .QINM ll mNH -. N. . .1 m n\u mmHUnHklx. -- . 3.0:n o.mw £06m 0.0: -.o o.moa o.mmm awe o I . .a :5, 30 GPfiaujnM Sm flHOD HQPHJ: 0.: pm amedaH n9: use sH 9.. .. a saa\ag L, Ar r. r jHOU H0962, .Hmnm “uh-ma .5. T). .| WEME MmpagmnOHQZ Z vfifiu MJSHJD mmaoho H3908 25+7H ?X fldde -57A- .CESAOU 03p 90 de deMHpsmocoo pa pcflom sedan a mo mmsaoop mHDwQOflpmozw myz *** -- -- -- m.mmm -- mam m Odom mmm mHomm :.ma o»w.a m.mm m.mwm .xma mm a.oa mum oxaa much w.m: mmw.o o.ama m.owm .mmm o o mmaa w» bah o.mm mm.a o.mw m.wwm mam no m.OH mom cam mwmm o.mm 0:.H m.m m.wnm Rm mw m.oa oma oawa mmam H.Hm 00:.H w.oma m.wum pm: ma m.oa mwa omw mmom m.mm mmm.a o.omw m.mw Ha» mm m.oa ooma mam mxmm*** 0.x: nmo.o 0.0m o.a~m 0.0H o o HHH Omma mam o.mm m»w.o o.am o.awm mm m@ m msa omma mmmm m .mm 95.0 0.2: 3% 5m 3. m 3E 8“ “:9 ~.>m H.H m.>dm o.a»m win mm m owna mqm moamx** mwzodH mag #50 CH 930 mpssflz 88 Java hogs; 03 ::m mmBI 44.80 $933 «89 8.35 aopflgmnmaoz 2 0:00 opddnm moaomo Hdpou nmgdm Buim aflhwvfidfio> p% zo\na ».Hm weHmzmm omHmo¢m UZHzodm amqeemm qqmz maH: 223400 :wma.m awe zo mzsm ozHaooqm mamasma: 92¢ mmqum Awmscflpuoov H>x mqm¢a =m>.om rmuHm: mZHmo¢m .wcHMHdooo umamcwmp mmda on I: w .mcflxodm on msflmpcoo cssaoo u: * .xpfiooam> mafidooam BOHwQ an m - 58 - msa. m.mm mam m:m o mma o.m zmma wdm mmaam* Hma. :.mm mom :pm 0 o 0 »:ma 3pm :aam* II II II mufim II 0mm 0 O mfiw OON .mmqrm -- -- .. mam -- 0mm 0 0 0mm m3: mmaam -- -- -- mam -- omw o o mma Ana «Adam -- -- -- mam -- 0mm 0 o Nam mam woaam :m.m wa.m m.ma Pom ma: 0 mam o.m on Hmm mmamm mm.ma mw.a m.:m m.wom mo: 0 “ma o.m :mma Pmm mmoam ma.oa so.m o.mm m.wom PF: 0 mma o.m mam HHm mwoam o>.m mm.m >.mm m.wom mmm o mma o.m amm Hm mwoam ow.» 0.: m.o: mam m.m»: 0 mad m.oa mm» 0:: mwoam ow.m cm.m o.m: mam :m: o mma m.oa Nam mam m¢moam mw.w m:.m m.m: mam He: 0 mma m.oa Nam mam mmoam om.m H~.m v.0m mam mm: o mma m.oa mmw adj mjoam mm.m :m.m o.mm mam mm: o mma m.oa mmoa mam mmoam p.m mm.m n.3H mam 0:: o mma m.oa m:m mmm m¢moam am.m am.m o.ma mam :H: o mma m.oa mjm 0mm mmoam mm.m mm.: m.@: mam m:m o mma m.oa axm mam m4aoam w:.w mm.m o.mm mam mm: o mma m.oa 0mm Ham maoam mm.m om.m o.mw mam ow; o mma m.oa omma o:m mooam ma.aa m».m 0.1% mam a»: o mma m.oa mwma mow mmmm mmaoaH mez pso cH use 2H mpznflz as I:8 umpmz oz cam mama 3H00 “@963 “mm mmasm sfiamHs umpflw\wmaoz 2 0:00 opsaom mmdoho H6908 mmpwm 30Hh ownpmazao> pg so\na ~.Hm “hpfimcmn mcflxomm monocfl-m».0m «pamfimm mawxowm wZon¢m mmgaemm qqm2,maH3 zzpqoo mozH->wH.m mme zo ma:m.oz~oooqm mmmqamz: mz¢memqu HH>N MAM¢B '4 -59.. the CClu as in the previous runs but no mass transfer could take place because the two phases were in equilibrium. The results, also listed in Table XVII, show that the throughput rates are very low; apparently mass transfer has to occur to obtain high throughput rates. A great deal has been written about end-effects in liquid-liquid extraction columns. Scheibel and Frey (23), say in the Encyclopedia of Chemical Technology that considerable transfer takes place at the top and the bottom of the column, and that if the height of the column is doubled, the number of theoretical stages may not double. Spargers or entrance nozzles apparently play an important part in the efficiency of some columns. To determine if the were true for the apparatus used in this thesis, a short column of the same diameter was set up without packing so the ends of the inlet tubes were five inches apart. Two runs were then made, one unpulsed and the other with a pulse amplitude of five millimeters and a frequency of 125 RPM. The results of these experiments appear in Table XVII and show that Simple inlet tubes, such as were used in these experiments, are the eQuivalent of only a small fraction of a transfer stage. Large Column Operation One of the main purposes of this investigation was to determine what factors affect scale-up. With this in mind, a 3.32-inch diameter column was designed and built in order to compare its operation with that of the 2.127-inch column. The 8-millimeter packing was added to the large column in exactly the same manner as used for the small columns. The height of the unsettled packing was 107.75 inches. - 6o - Expanded End Sections The large column differed from those which preceded it, not only in height and diameter, but also in that it was constructed with special end sections. These end sections should in no way affect the efficiency of the column, since the inlet tubes still terminated at the packing supports and the two liquids did not remain in contact any longer because of the end sections. The end sections were added to give more settling time at the outlets and, if necessary, to eliminate the effects described by Blanding and Elgin. One of the disadvantages of expanded end sections for experimental work became apparent after the first few runs. As noted earlier in this report, the interfaces were maintained two inches below the bottom packing support and two inches above the top packing support; however, it made very little difference in HETS values or throughputs whether this was two inches or four inches. With expanded end sections present, this two inches brought the interfaces down into a section with a large cross sectional area. If the interface levels were changing slightly while samples were being taken, the rates measured would not be a correct measure of the flow through the packing. In order to overcome this, great care had to be taken to be sure the interfaces did not move during the sampling. Since the end sections were made of light glass they were very fragile. Several were broken during the course of this investigation. Probably the most important disadvantage of these expanded end sections for experimental work was that they held a large volume of slow moving liquid. Because of this, the column had to be operated much longer at steady state conditions to be sure that all of the . ' - . ,_. - 61 - liquid in these sections represented exactly the liquid coming out of the packing. Unpulsed Runs on Loosely Settled Packing» Table XVIII lists the results of unpulsed runs using loosely settled packing on the 3.32-inch column. All of these experiments used carbon tetrachloride feed containing 1% acetone. In the initial runs, F118 through F123, the interfaces were regulated in the expanded end sections, and the rates may therefore be somewhat questionable. This was confirmed by poor material balances in these runs. For all runs made later, the interfaces were maintained in the narrow section of the column. Unpulsed Runs on Well Settled Packing The packing was settled by pulsing the carbon tetrachloride in the column for 12 hours with the pulse amplitude set at 10 millimeters and the frequency at 65 RPM. At the end of this time the packing height had decreased 9.75 inches from the original 107.75 inches. Since the packing should have settled more than this, the frequency was turned up to 125 RPM and the pulsator allowed to run for two more hours. The packing settled 3.5 inches more, and again the frequency was turned down to 65 RPM for two hours. The overall change in height was 13.75 inches. The results of unpulsed runs on the 3.32-inch column, using settled packing and having a packing height of 108.25 inches, are listed in Table XIX. It should be pointed out that, at the beginning of this series, the entrance tube for the carbon tetrachloride broke off just above the bottom rubber stopper. Since it was felt that this should make little difference in the results, it was not repaired until the next series of runs, the results of which are reported in Table XX. .meHpomm duo map mm m: 6H0: 09 0:6 muons g has mane usfldeeoemm sped mpdh amassedm .meHpoom who sodsdmxm 0:9 as snap Heaven compomm wound: 039 sq dopsfldmoa was momHMOHCw exp .96 many mopws meme mama mom I m r m.am ma.a om.:m o.ma 0.1mm m.mmm omam mama swam m.sm mm.a 0m.3a 0.9ma o.amm Ham “mam pm mafia -- -- ow.me -- m.mwm ea New comm emu. -- .. ow.me -- m.mam ~.m meg ommm swag -- -- oa.mw -- m.mmm m: rs owmm made . m.sm mw.a mo.aw 0.:H m.mwm oma moa ommm soda ,mw 0.00 ma.a om.am o.ow o.»0m mxm 00am omaa mafia . m.m> ew:.a oh.mm 0.0» o.wom Hem oasm omam wage :.as mam.a om.am m.as o.aom :wa mama mmmm amaa o.sm mmw.a 0n.me m.m m.mmm em am: owes mam ms.mw sar.a oa.eo o.eaa m.omm was omem mesa mafia o.ms was.a mm.aw o.pm m.mmm :ma mama oosm mu Ha -- -- oo.aa m.m o.»am we see mama amass m.mm mo.m e.m_ o.m o.eam ow Hms mmmm mesa mozomH mHz mpoom wm pmo CH 930 :Hoo Hops: oz cam mam: mo 85m :HOQ mops: m«:\as aowmqwnono: Z 0Com mnzaom momma soak ofipwofisao> as sU\sa m.e: saHmema ozHeoam :ma.soa aeoHa: oaaaoem UZHeoem oaqaramss eaHz zzsaoo :mm.m so mesa oaHeooaa mamasme: HHH>N HAM¢B . :x o m.mmm was com mama mm a -- mm.ms o m.mmm mm Hma ema emam mus.a cm.m~ m.mw m.mmm mam oosa Hmma emmaa : was.a os.me o.mm m.mmm omm oo:a mama mmaa aw NMm.a _c.m> 0.0a m.mmm mmm mesa osma mmaa we mom.a wo.a. m.mma 0.90m ms: ocem mam Hmaa He can.a m.Ha m.moa o.m0m 0am owam oama Oman we: msoom an 930 eH use :Hoo amps: on sea .Ho 33m IIHUQ .3983 mfiKde. Howflflmnoaoz III 330 3: ..m no)dm be m “Humorndés V! mammommm .3: mgm Broad. 9me fiomfimm KJBQZ OI HEM...» PH 50\QH ®.m: M. HQHQ OWN/Iowa :mm. OOH EIUHEI UBHVIU>.a maa o.mam n.ema o 0 mm m pom smaa m.mm nma.a ama m.mmm o.amm o o comm aom mmaa monoca me: use ea ago apnea: as same hope: on use mew: :Hoo Hops: hem muadm sas\aa hemag\nmacz E 0:00 mmsHoa moaumo Admom noam; seam mappmesao> P“ 50\QH :.Om MaHmZMQ OZondm :O.HOH_HEOHmm UZHKUdm OZondm QHQBBMQ mBHZ ZEDQOO :Nm.m ZO mZDm UZHQOOQM QMmADmZD 02¢ QMMADM xx Mdmda \ “ . - b .maasoazfsau dodaiaopod ma: In * ma.“ m.ma 0:.m 0.xmm o.nmg we A.m amm am: IaImmmaa mn.n m.ma ma.m 0.1mm 0.:mm my a.@ yam pm; mgaaa ~m.m m.>a m.ms m.w a A.Imm aw x.a mam was Iavaamaax Im.m m.aa m.:w n.waa m.mmw mu A.m mam an: munaas :..m a.ma :.oa A.ca1 0.,Im an oa.m emoa mum mafia. ma.c :.ea n.ua 9.3on o.mou xma an.“ arm a-m Iavaawaa :e.a m.»a m.aa o.sma «.mem nma em.m mam am: m-.aa em.» a.aa m.ma m..um 0.:I. mu afl.m mmoa oak Iavmm as wo:Q:H 29: p53 . 2H as; onzsmI Eu 1 03 page: on cam name :aua guns; sum mnasa maeIa: ssaaameoa.; a ;:;m unease soauso aana so_sm :;aa maesogxan> mu :0\ 3 r CZHZU - N. 344.: ...xJJ... l . n n . h l _ ea . t u ..l4J .tx? . .1 l L ..m has Lt Pet's .J.h ‘ . l n. : | A_ a 1 -91.. I t l 1 Sn; " .1 N ', . . ”w ——-‘kH~.—g.‘-—n a i I l E 1 g 6- .. f . k ' -.....r.‘-._4 . ' 1 1 I ..l"l V ‘ tint}. , .3_ . . , ,. .. .. .. . . . . .. . . .. _ .w. .. .0. w... , .. _ .... o v . , .. . . ..:.. o . O . V’I'IIlllv".l J 9 I I _L._ Q U a- . .— I l T" T.“ 5 J l I ’ Mt. l I ho... 1~——o~—q I t f .L.-.Ll.l_...._..'.-_ '5 - 92 - in critical flow ratio with an increase in column efficiency; that is, the higher the frequency the higher the efficiency and the higher the critical flow ratio. In this series of curves the flow ratios vary from 0.8h3 for the minimum frequency of 65 RPM to 2.h2 for the maximum frequency of 215 RPM. Particle Dispersion When the first pulsed runs were attempted in this investigation with settled packing, the contents of the column turned white and milky. Throughput rates were greatly decreased, and difficulty was encountered in separating the two phases in the end sections before they left the column. This phenomenon was less pronounced at high throughputs than at low. It was also less pro- nounced at low column efficiencies than when the column had a great many theoretical stages. The investigator thought that this was emulsification because a great many references have been made in the literature to emulsification occurring in pulsed column studies. Considerable effort was expended to eliminate this phenomenon. Experiments were made on a Waring Blender to find out what could be causing it. All of the carbon tetrachloride was steam distilled because it was thought that the trouble may have been due to contaminants in the organic feed. Even the feed water was treated to remove impurities which might be present. Nothing that was done seemed to offer any solution to the problem, so the experiments were continued with what was thought to be emulsification. One of the convenient ways to make a column operate more efficiently at given concentrations of the solute in the entering stream is to increase the amount of mass transfer occurring within the column. A.greater amount of interfacial area has to be Obtained -93.. to do this, and to get more interfacial area the liquids must be broken up into fine dispersions. The more efficiently the column operates, the finer are these dispersions. The author has reached the conclusion that what was originally thought to be emulsification was actually only fine dispersions, and these are necessary for the efficient operation of a column, even though they result in reduced throughput. Factors Which Influence Column Efficiencies Throughput Rates The choice made for the method of column operation in this investigation has proved to be fortunate, not only for determining throughputs, but also for determining column efficiencies. The column was always operated at maximum throughput rates when column efficiencies were determined, except in a very few runs where throughputs were intentionally decreased to see What effect this might have on the efficiencies. In these special runs, the interface was regulated at the middle of the column. One of the reasons that this method of column operation was fortunate is that it eliminated the necessity of making a choice of which phase to make continuous. Three interfaces were always present, one at either end of the column and one in the side tube. The inter- face in the side tube automatically moved up or down the column with changes in flow ratio. If a large fraction of carbon tetrachloride was being fed to the column, the middle interface moved down so that water was the discontinuous phase over a greater portion of the column, and vice versa. Actually at flooding the two phases are indistinguishable and it is probably incorrect to speak of a continuous or discontinuous phase when three interfaces are present. A .e,‘ ‘- nd‘ " J. J- .o.\,. Fr! ,A Fri The method of balanced column operation was also advantageous for obtaining column efficiencies because it eliminated the effect of decreased throughputs. .A series of runs were made on the 2.062-inch column in which the carbon tetrachloride rates were kept constant while the water rates were varied from flooding down to 10.3 (ft/hr)l/2. These runs were not pulsed. The data originally listed in Table XII have been corrected for entrance effects and are repeated in Table XXIII below. These data are also presented graphically in Figure 17. TABLE XXIII EFFECT OF REDUCED THROUGBIUT ON NETS (UNPULSED) 2.06-INCH DIAMETER COLUMN Flow Rate EEHS (cu ft/hr/sq ft) Sum 1 (Inches) Run No Water CClu (cu ft7‘ sq ft) NTS Corrected* Rot 25.65 26.1 10.28 0.828 h5.0 R65 20.10 23.25 10.27 0.775 #8.8 R66 76.50 22.t 13.u8 0.576 71.6 R67 13u.2 22.7 16.36 0.t52 101.0 R68 u9.1 22.15 11.72 0.666 59.1 R69 92.0 22.0 1u.29 0.535 79.2 Flooding 170.0 30.2 18.52 0.525 81.3 * These values have been corrected for entrance effects. The graphs show that NETS values are greatly affected by the total amount of the two phases through the column. In other words, if the column had been operated at any fraction of the flooding velocity, then all of the NETS values obtained would have had to be corrected for the change in efficiency due to this reduction in throughput rates. {.3 gr: w 01.)». ..1v. - . -, -- 3.3: m3 MCDV . .. 911314 Urx u UryoML. atlmmuusuwmrwwlili. .1;1+NI\1\N \ .. .n- 347.-.}; 2 .1.» ..-.ImwttlmW... . ‘1‘: a. e. ..n. 2:15.». .-. . \ l , \I .1 \ u s . . 7. n o o O a a a a ,i o o o a o a I J J V I, J g a a o a o u 9 o c ,_ __ u L o c o o 0 v u 1 u v .. O T _ . I O Q . I O C » CU - 96 _ Figure 17 indicates that an unpulsed column should be operated either very close to flooding or below 75% of this value. The HETS values become sharply poorer just below flooding, reach a maximum between 85 and 90% of maximum throughput and then start gradually improving, so that at 77.5% of maximum throughput the mans is the same as that obtained at flooding. Below 77.5% of the maximum rate the HETS values gradually keep improving without any limit that could be found in these experiments. These results should be observed with some caution, however, because a large difference in operating conditions occurs between flooding (3 interfaces) and nonflooding (1 interface). Garner et al (29), say that unpulsed packed columns should be designed near the loading velocity where the HETS is a minimum. A series of runs was also made on the 2.127-inch column to determine the effect of reduced throughput with pulsation. The frequency was set at 125 RPM, and two different amplitudes were used. The CClu/water flow ratio was kept approximately constant at 2.1. These data are given in Table XVII. The sum of the volumetric velocities of the two phases, which represents some fraction of the maximum throughput of the column, has been converted to a per cent of the total throughput in Table XXIII (a). -97- TABLE XXIII (a) EFFECT OF REDUCED THROUGHPUT (PULSED AT 125 RPM) 10.5-mm amplitude Water Flow CClu Flow % of Maximum NETS Run N0 ml/min ml/min Throughput Inches F99P 706 1375 100.0 11.15 FlOOP 6&0 1220 88.5 9.62 R103? 583 1096 79.8 9.22 R105? 515 887 68.7 8.82 R105? (A) 515 887 68.7 8.80 R106? tho 768 59.0 7.70 R102? 250 3A3 u3.8 6.26 R102? (A) 235 3A3 28.6 5.76 5.5-mm amplitude FlO9P 657 125h 100.0 16.23 R108? 311 616 u8.2 10.12 R107? 91 221 15.7 3.70 These data are presented graphically in Figure 18. They show that the efficiency of a column is dependent on the total throughput rate. An unpulused column is much more dependent on throughput rate than a pulsed column. Pulsed columns show a maximum NETS at flooding and this gradually improves with decrease in throughput. The higher the pulse frequency and amplitude the less dependent NETS values are on throughput. From this it would appear that pulsed columns should not be designed to operate near flooding. Cohen and Beyer (7), working with a pulsed sieve-plate column, reported that NETS values are fairly insensitive to changes in flow rates at higher pulse frequency, and at lower frequency the HETS varies over somewhat wider ranges. Chantry, von Berg,and.Wiegandt (6), state that lower stage heights are expected with increased rates as A 4 k- . 1. . , Cr .: a. I 53.... Rat—X” 1.l 1 1 1 .1 fl a M m . 1 a 1 W . a . .Jtzfl1121fi1%1:711:- )1 .. .1. n .. r IrtvIw-fi ,1 . . it i- - 91 . -I It llr-yb . - .. . u . _ 1. n _ V 1. n m . .... m .. 1 . 1 . iii. . _ 1 ...--... m -....fidmamlo -§§-Mo1.nm ohm. N»..- i. 3.1:: {...-«..., 33...:- i--.--: *1 ”J .77 1‘ T J 1.- 1.... - v . - u | ..q .. 11....» . -.....— - x 0 _. 1 E T 1 I I I 9 O r 1...,__..1,1_ . L~--T-—»-.-—-~J—- -4 . . ’- . . " ‘ ' ’ l , . i 1 I a 1 1‘ 111 - ! »-1-Tl “I .‘1 i J 11 ”H1L-1 i 4 f .L i l J ’f"‘ 1 ~‘ ., . .n .... .. v. . A: . I. . 1 1 11 5 Ta. i J ' i I i i L i l i l l i 3 "T 1 L1 4 E h‘f‘ f 1 7 l1-1,1 If .1f \\ "; L l 5 1 1 l ‘i L f i i1. 1. I ‘ is £111 1-- . 1.. ¢ . ...... . .Q .... h w .h . o '. . . ... v a .. . ( . . .- ~ 0w 'L V . ... v H w . .. - auo.fioo.v . .E... ..... . . . k .. . . v .. . .. v 1.0 A f. E -h111l1. i , :;‘ ..! g : ”“T‘ 1 1 1111,--- 11 ..L 11 i - . . . .. . . . - l .- . . a . . . . . 1. :- T- .----..};- - . ti. . _ . . E1... -. rfl..- .-.-«4.1. .. . .- Lint -111 l 1. $ 1' .1 W l... A”..- ..L _. F f 4— --L --— -. 1;-)r1 L . 2.1-.11-1.---f---;1...1-..._ - , - L ' 1-11.. I 1 l i ? “.111 -J_ - > +"1Trfl". T f l -11 F ‘- 11 1 11-11 i 1 l l j . P 1 5 1 1 1 t 1 f l f I I l + : 1' 1411,- I __1r. 1 -- L11- 1 t t b l 1 -1 1_. f L I nun—1 - M 1.11 * ,-.-,---.a-.11m.§om Mam-wag 116-assess; 4.1318. x , twat. 3.6 t a at w . 1E axe-1.1111......-assess... s1. $11513.- sum-i. ems-.11... Elsa-mags a . 1 . a . l- . . -. .1 , . . - , w ‘ . . 1 _ v «s... - ._ .. 1 1 . .1 1. h . 1..-..- 1 a. Bows mm .m «Emma # 0 H THE-«rm -99.. a result of great turbulence in both phases, and that this effect should be less noticable on a pulsed column because much of the turbulence is supplied from an external source. Thornton (lb), working with a pulsed sieve-plate column, also agrees with the results obtained here. Flow Ratio The concept of a theoretical stage has been used for many years as a convenient method for designing distillation columns. In this theory, the column is considered to consist of a finite number of equilibrium stages. The usual plate distillation column actually does consist of a finite number of contact stages or plates, but the plates do not represent equilibrium stages be— cause equilibrium is rarely reached. The process is a countercurrent stagewise system but certainly not a countercurrent equilibrium stagewise system. Objections have been raised to the use of the theoretical stage concept on packed columns, because packed columns are true counter- current processes. A packed column does not consist of a finite number of equilibrium stages, or even a finite number of contact stages; because of this, the HETS concept is not recommended by some authors. The HTU theory, on the other hand, has been proposed to replace the theoretical stage method. This is defined by an integral and can treat the countercurrent process as it actually exists in the column, in a differential manner. The HTU concept assumes equilibrium at the interfaces; this is a highly complicated phenomenon which must rely on instaneous mass transfer in these areas. To find the number of transfer units, one nub-r Irv-- a boy r _,. .- ..L‘... c- {’1 (IV (1. . - lOO - must assume that either one film or the other is controllin and that the remaining film contributes nothing to the resistance to mass transfer. In practice, both films contribute different amounts to the overall resistance and neither has a tendency to approach negligible proportions. In the final analysis, any method has to be judged by the results that are obtained by using it. For example, a packed column operated in any given manner should have an equivalent of a certain number of theoretical stages. This number of stages can be thought of as some number of plates in a distillation column. The number of plates cannot change regardless of the flow ratio. That is, some of the plates are not discarded because the flow ratio changes. The same is true of the inlet concentrations of the two phases. Just because the entering phases might have different concentrations at one time than another does not mean that part of the plates will disappear. On the other hand, if the operating conditions change, the efficiency of the plates should change and the number of theoretical stages will be different. A series of unpulsed flooding runs were made on the 2.062-inch column using non-settled packing. The carbon tetrachloride-to-water flow ratio was varied from 7.h5 to 0.178. The concentration of acetone in the entering carbon tetrachloride was varied from 186.7 to 265 millimoles per liter. These data are from Table XII, and are converted to HTU values in Table XXIV. The subscript 0A refers to overall mass transfer based on the aqueous phase, while the subscript OO refers to overall mass transfer based on the organic phase. n... a... fir. fir; 51.. «...-L ma. .0. VI. a... \ F...- F t- - lOl - TABLE XXIV INFLUENCE OF FLow RATIO 0N HETS AND HTU 2.062-INCN COLUMN NON—SETTLED PACKING Flow Ratio HETS HTUOA fflll Run No CClg/HQO (Inches) (Inches) (Incggs) F56 2.10 61.7 61 59.8 F57 3.18 66.3 5u.6 81.5 F58 5.30 57.6 37.8 9u.0 F59 7.u5 56.0 31.9 111.0 F60 1.10 6u.5 91.5 u7.6 F62 0.u81 65.9 152.0 3t.u F63A 0.178 5t.u 2u1.0 20.1 These data are presented graphically in Figure 19 and clearly show that HTU values vary over a wide range with changes in flow ratio, while the HETS values remain essentially constant. As further proof that HETS values are independent of flow ratio when the column is operated at balanced flooding conditions, other runs were made using settled packing, both pulsed and unpulsed. These data have been calculated in both HTU and HETS units and appear in Table XXV below. TABLE XXV INFLUENCE OF FLOW RATIO ON HETS, HTU 0’ AND HTU 2.127-INCH COLUMN; RACKINC DENSITY, 51.7 lb/cu ft; PULSED Ah UNPULSED -—~ Eulsed at 10.5 mm and 55 RPM Unpulsed Flow Flow __hun No Ratio HETS HTUOA HTUOO Run No Ratio NETS HTUOA_ HTUOO F9h? 8.85 22.5 11.78 u8.7 F80h 1.7 to 50.1 to F95P 1.80 21.1 2u.0 20.2 F93 0.0575 uh.6 uu.5 11.2 F96P 0.07u5 22.0 182.0 6.35 F97 1.h1 A5.8 55.3 36.6 F97? 1.1h 22.6 30.7 16.u0 F76 0.081 no.7 3u.6 13.1 F70 6.2a M5.7 27.3 79.6 The data above are presented graphically in Figure 20. - _ M 1 _ . ‘ aaflafielsoah a: C» 2.: - ._ 1 an HANNA... Ina...“ Hui-AN §3éj.fl..wN--T Q. (zmvw-WNVEMEVWV M1 .. Jail-«SNWHI.Tang-Znaljflafim ES... ...! €7.11. .- 5 - a . \ -.. r. a -Ll .le 8-14TH-n1he \ bit-:- fnaaovwngtnyfi ..fim-k..lCm+-WI£1til-11-1.s111l4L 1 1 1.a.-s111_._1.1 . 1 .— a all... 1 -103- - 10h - Even though NETS values are relatively insensitive to changes in flow ratios when the column is operated at balanced flooding conditions, an obvious question is whether or not the same is true of columns operating below flooding. In order to answer this question, a great many literature references containing experimental HTU values plotted against flow ratios have been examined. The HTU values based on the phase not already appearing on the graphs were added. The HETS values were also added to the same graphs. All of the plots examined showed that HETS changed much less than HTU. Two graphs were selected at random from those observed in the literature to illustrate the point made above. These data are tabulated in Table XXVI below. TABLE XXVI INFLUENCE OF FLOW RATIO 0N HETS, HTUOC, AND HTUOD Taken from Thornton (1D)i Taken from Treybal (3Q) UC/mUD HTUOC HTUOD HETS mUD/UC HTUOC HTUOD HETS O 0.8 2 30 60 u1.5 1 1.6 1.6 1.60 h 17 68 31.u 2 2.32 1.17 1.62 10 8.u 8h 21.5 3 3.0 1.00 1.65 20 u.9 98 15.h u 3.8 0.95 1.76 to 3.05 122 11.7 5 h.6 0.92 1.85 60 2.2 132 9.1 80 1.79 1&3 7.8 100 1.5N 15h 7.25 The data from the table above are presented graphically in Figure 21. It can further be proven mathematically that the HETS values must always lie between the values of HTUOA and HTUOD by the following relationships which were developed to be used with a straight equilibrium 13...... 1.3... .1... ...-...... ...-....-wT . 3... o o o a I o . a 1 _ x o 1 I c n o u o .1 \ O D O I I . n x J a 0 I I O 9 O 1 \ 1 1 _ _ I ,. . + . . _ . i "'—T“"’ ...b\ 9!. Iggy V“ be (i - 106 - line and operating line. HTU =HETS(l-P)=HI’U 1 01“ lnl/P 00 P where P = m L/G m = distribution coefficient L/G = flow ratio Let m, L, and G take any finite positive values, and substitute these into the equation above. The values of HETS will always be between HTUOA and HTUOO. By mathematical reasoning, the curve representing HETS values must have a slope less than or equal to whichever HTU curve has the maximum slope. The boundary conditions are such that the HETS line can never cross either of the other lines. Therefore the HETS values must change by an amount equal to or less than the maximum change in HTU values. The conclusion reached by this author is that HTU values are much too erratic. They are highly dependent on flow ratio, the correct choice of which film is controlling, and other factors. Most import- ant, HTU values are highly dependent on which phase is continuous and which is discontinuous. End Effects Two runs were made on the 2.127-inch column to determine what fraction of the total mass transfer occurring in the column could be attributed to the entrance and exit tubes. To do this, the packing was taken out of the column so that only the lines carrying the two phases into and out of the column were left. Run Rllh was made without pulse and RllSP was pulsed at 125 RPM and S-millimeter amplitude. The results which appear in Table XVII show that only O.lh of a theoretical stage is due to the entrance and exit effects. ... wo' o- (‘n -107- Sherwood and Pigford (31) say that the HTU values in some columns may vary by as much as the first power of the column height. This effect is particularly noticeable in spray columns but has also been observed in packed towers. Murch (ll) has developed an empirical formula for calculating the HETS for a distillation column. In this equation he suggests that ms is proportional to the 1/3 power of the column height. A great many other investigators have reported similar results on studies of experimental columns. The conclusion reached here is that end effects are not important unless special spargers or distributing weirs are used. The small amount of extraction due to end effect remains fairly constant with or without pulsation, and allowances can therefore be made for it in interpreting experimental results. Orientation of Packing Some experiments in this investigation exhibited an effect referred to as orientation, for lack of a more precise word to describe it. This phenomenon was usually apparent in the first series of runs made after the packing had been originally placed in the column, but was also noticeable to a lesser extent for the beginning runs made after the packing density had been changed. Illustrative of this point are runs chosen from Table XIII using well settled packing on the 2.127-inch column and repeated for convenience in Table XXVII. In the same table runs are listed for the 3.32-inch column using non-settled packing. It should be pointed out that after five to 15 runs had been made on the packing, the results became reproducible and no further changes in HETS values due to this phenomenon were observed. ##7” - 108 - TABLE XXVII EFFECT OF PACKING ORIENTATION ON HETS FLOODING RUNS ON THE 2.127-INCH COLUMN PACKING DENSITY 51.7 lb/ft (unoriented packing) Unpulsed Pulsed at 5 mm, 125 RPM Sum of The Sum of The Run N0 HETS Square Roots Run No TETS Square Roots F71 90-5 57-93 FYOP 29-7 65-80 F72 75.0 58.50 F71? 25.5 65.5% F73 67.8 56.50 F72? 20.5 6u.0 F7h 68.2 58.10 F73? 17.2 61.7 F75 50.5 5h.2O F76 ”6.7 55-19 FLOODING RUNS ON THE 3.32-INCH COLUMN PACKING DENSITY uh.5 lb/ft3 (unoriented packing) Sum of The Run No HETS Square Roots FlEI 77.h 9h.80 F122 73.5 92.90 F123 60.6 97.20 Fl2h 57.8 --- F128 5h.2 9h.30 This phenomenon was not observed every time a change was made in the packing density, nor did it seem to affect flooding rates on any of the runs where it was observed. Similar results have been reported by other investigators. Thornton (lh), for example, working on a 6-inch diameter packed column, concluded that packings tend to orient, an effect which leads to a progressive increase in over-all hTU and a decrease in throughput. -109- This author can only agree with the first part of Thornton's statement, that packing tends to orient. However, it is obvious from Table XXVII that the efficiency of the column increases, rather than decreases, with this orientation. Furthermore, the limiting throughput does not show an increase but appears to remain essentially constant within the experimental error expected for such unstable column conditions. Packing Density The packing was poured into the top of the column which had previously been filled with water. When this was done on the 2.026-inch column, the resulting density was #7.1 lb/cu ft, while the 3.32-inch column gave a density of 4h.5 lb/cu ft using the same procedure. Apparently the speed of adding the packing had some effect on the resulting packing density. The speed of pouring the packing into the column was never uniform. A.series of flooding runs was made on the columns to see what effect the non-settled packing might have on column efficiency. The packing was then settled by pulsing and another series made to see what effect settled packing may have on HETS. The results are tabulated in Table XXVIII. *« - llO - TABLE XXVIII EFFECT OF PACKING DENSITY 0N HETS AT FLOODING (UNPULSED) 2.0265-INCH COLUMN Packing Density h7.l lb/cu ft 3.32-INCH COLUHH Packing Density 50.1 ih/cu ft 3.32-INCH COLUMN Packing Density hh.5 lb/cu ft 2.127-INCH COLUMN Packing Density 51.7 lecu ft Run No EETS Run Ho HETS Run No HFTs Run No T’TS F56 61.7 F70 h5.7 F118 52.3 F136 58.2 F57 66.3 F72 75.0 F1188 73.0 F137 57.1 F58 57.6 F73 67.8 F119 62.75 F141 57.6 F59 56.0 F7h 68.2 F120 57.0 F1u2 58.u F60 6u.5 F75 50.5 F121 77.4 Flh2A 58.2 F61 52.0 F76 #6.7 F122 73.5 F6lA 59.0 F80h no.0 F123 60.6 F62 65.9 F81 17.u F12u 57.8 F63 19.2 F85 37.6 F128 51.2 F63A 5h.h F89 38.0 F129 61.2 F93 uu.6 F97 h5.8 .Average 58.66 Average 50.61 Average 62.97 Average 57.90 *Corrected 82.h0 *Corrected 67.20 *Corrected 69.0 *Corrected 63.h0 * These values are corrected for end effects. The data in Table XXVIII are presented graphically in Figure 22. The curves show that the small column gives a much greater improvement in efficiency than the large column for the same increase in packing density. This might be accounted for by the fact that packing in the small column had a great deal more tendency to orient than packing in the large column. That is, a large portion of the packing in the small column was either vertical or horizontal, positions which are expected to give the most efficient operation. Very little of this type of packing alignment was noticed in the la‘ge column. The author has been unable to find any reference in the literature to a qualitative study of the effect of packing density. There have been a few references to the phenomena of packing settling, or orientation, but only in a general way. . r T a :M. ...ZTI...1.V.. «m ....(1 . c. . a . _ o o 1 _ . _ . . 1. . e v 1 e 1 o o a o . . . . 1 i 1 1 u . _ _ . C O . . . . 1 1 . _ _ 1 0 o -1 . if. 0...? ..- T~ ~, CI «.1- u 'meurme. ulJmeLWd‘MM...W~N ..1Nvax1.¥111 Sang... .7u...31l!fl.‘. . 1. .e N z 1 0 ./ a . l 1 l a . J v y a . I I I \ . 1. 1 H1 \1 . .. 1 o 1 ; q 1 1 . 0 .. 1 . . g 1. A J 1 g I 1 . J , I a I . a 1 ... 1 _ I JI! F J‘J}~iui’. I. I:__ 1.: K. n...- . 1‘ ll.‘lll|.4. I a _ 1 t 019-9.-.1II - 112 - Pulse Amplitude Several series of runs were made to determine what affect amplitude would have on the operating efficiency of a column. The first series was made on the 2.127-inch column with a packing density of 5l.7 lb/cu ft. The frequency was held constant at 65 RPM. HETS values were determined at two different flow ratios to be certain that changes in flow ratio would not change the column efficiencies. The results of these experiments are tabulated in Table XXIX. TABLE XXIX EFFECT OF AMPLITUDE 0N HETS AT FLOODING (PULSED) Frequency 65 RPM Packing Density 5l.7 lb/cu ft Packing Height 30.75-in CClg7H20 Constant CClu Rate Flow Ratio 2.1 llO mlzmin ‘Run No Amplitude NETS Run No Amplitude HETS -- 0 11.5 -- 0 ul.u F9lP 2.0 32.5 F92P 2.0 35.0 F86P 5.0 26.1 F81P 5.0 26.1 F95P 10.5 21.1 F96P 10.5 22.0 These data are presented graphically in Figure 23 and once more show that HETS values are independent of flow ratios. The curves show a gradual decrease in HETS with increasing amplitude. Another series of experiments was tried on the 3.32-inch column to determine if amplitude would cause similar decreases in HETS of this tower. The packing density was 50.h lb/cu ft and the carbon tetrachloride-to-water flow ratio was maintained as nearly as possible at 2.0. It should be pointed out that the reason for the close adJustment in flow ratio was to try to keep the operating line and -113- I. ...... . ibu— I 1 q . 1 11 .III 1 ..1 _ i J . II III I4.-I.4 .414IIII - IIIIII .1 I- 1 11 1 11 1 11 . 1 .1 1 1 1 1 _ 1 1 .41 b 1.. . 1o ..t...III ‘I§_II I .II '44! 'l4 I I A" ..~IIII4 4004..“ OL 041.1 In! . c i . 1 . .. w 1 1. _ M __ M w m . 1 fikmngmumAWwbn- I‘M“ JWHwfi “1H5?“ “- 4 .1 .. I f I 441 .-- -w - 4 14I- II . . 1 j . . a 1 . _ 1 . 1 1 1 1 .1 1 1I 1 :I1 111.4 I. ... .1 1 a o 1 1r .1 v. 1 .. .1 . ..1 .. 1o ., . . N . 1 1 W 1 1 . .I . 1 1 11 l . q 1 . . 1 . . .1 . 111 . . . .. 1 . 1 .1 . 1 . 1 - LIFT--14 III-I“. ,I 4- 1- Elly-III . - ..1 LI 1. 1 1 1 u 1. 1 m .1 . . 1 _ Il- . . 1. .. I- . I- If - 41-».-- fi-I. 1- I ----.1. :1 11:51 ---.II; - . I-”- --.1:--. I - .1. - 1-4.1:- - --_-.I 1 1 .1 1 1 . . . 1 . 1 . . 1 1 1 1 1 1 . . 1. 1 1. a 1 .1} -. 11 . 1 t a 1 1 I1 1 . 1 1 u. . 1 1. 1. . 1 . . _ . . H .1. .1l M t w « .1.| > L1 1 1 1: 1 . 1. :1 . . 1 1 1 W .1 1 1 _ . m _ A. 1 11 .1. 1 I. . . .1 . 1 ._ . . . . .1 1 . .. H j“ w 1 . .. . 1. -II ..... -1112... -4 .I 411417-41. 11.4-4 1 -..“IIIIT. 1” w 4-----1-IIIIIII..-I-..T-III- 44 7.!“ 41-111.91-141 - 1.4 -......4- - I14 I1444-1II 4 4...-- I 1T ..1 1..- .1_.._fi. .1. 1.. 1. .1 . 1 .1 .1 11 11 .1 . y 1. w :1. . _. _. 1.. . 1.. 1. .H ... 1_ 1 .1 1. .. . 51. 13-1 1-: .1. III-,4- I- III - : II- IIII.11I- IIIHTII1 I :1 4411-414, I :11 II+- 1131-}-- 1 I II-11-I;-. -I+-.--114...I11I4 4- II 1,-1.14 .1 111171 .1. ._ 1. 11 1 .1 . . _ 11 11 1. 1 11 1 , 1. 1 . 1 1 1.1 J 11. 1. 1 . 1 .I. 1- - 1 1 i. .1 1 1 . .. 1 g .1 1 1 1, w .1 . 1 .1 T g . . 14 IIV 4- -. -4 Inf-44. 1 ...-I41. 44- I..I.I1TI.I-I. I+1I44 - .1 -11-,.4.I 74:11}. III... .11 . I1 ..... -4.4-114 . .-4-4.I4I-. I414... 144 . . m. 1 ...... 1 .. ..axr . . . 1...nII1 1 144.- .-I .I .I c D -1. .-.. 1 ‘10-...“ 4 W—JL m 4.1%; m4—o—m 1.__& —---—--—> n . - ....._ H-—-r——--wrh >¢—~o—~ .1 -1. 1 .11-- 7»- -_-_..11- -“.:‘_..— . I + 1 I I I I L ..1 w . . 1. -, m 1. . a . .a 1 1. . 1 . .- H .1 . . T ..-.1.Iv.1- -...- -I. 1 ----4 I41 I-1T4I .4-.I.. ..L41- .. I 44- . w . . I .1.II-.I ---..IIItI- . a i . .. . . P 11 . ...»...1 -. 1. It“ 1 1 . . 1.. :1 1. . . I. 1. I. 1... km .1 m 11%! I. . I. .. .1. I. . . I311! 1 . II. . . 1 w .W . ... . . .7, .1. 1. 1 j M . w M 1m , , . . 1 1 U ... . H . . .. 1 . 1 .. . . .. . . . . . . a: 1 41-- 1 I I ..-.I-.. . -..1--41r-41--.14 LII III-$1141: --IITI .. .4 -1 Pl III-I4 . .. .. 1 . . .. . . 1.. ... . . . .. . 1 1 :1. . ,1 . 591 1 1 . .1 1 .1 1. :1. .1 1, 1. .- 1 1. 1.1.“ .u. .1_ I . .- . . u: . .. I1 LI. 11 1... III! 1 .. .1 1 . . 1 . 1. . . 1. 11 1; , 1 1 _ .11 .I .I 1 .II.» -I I 1. ,- . - 1 - .III14.I...IIII.114I..- 141.4-- IIL.1I.4I1III~.II-TI.1 w L 1.15111 .. I4 .11. I II..- .1144 I14. II11 III h14I4L 4. Ital-u!- 1 . 1 11 . _ 1. .1» .1 R1 1 . w 44IIFI|. II ..I 1. 1 P o .F . 1 II-IIIIII .1 2.1 . :1 .... . 1 r .1. 1 1. 1. 1 . / .1 .1 1 1 .. . . .. . , . , , .1 1 1 _ . .H 1 1 1 . f . 1. .-1.-II1+-, 3.19.11 “1 .4 w .. III??- - I- 1.3 4I-I144I 1.1.1.4 I4441.-III.- . Z 1 . . -1. a . _ 14-1. 1 1 1 .1 ..1. 1. ._ 1 .1. .1 1 1- 1 1 w .1 .1 .1 I. . 1. 1 .. I. 11 I14 1. .1. , ...OmL. III-IL-I III-III. II.1- 1. 1 1 , 1 1 . . 1 1 1 . 1 .. 1 *gv II“. . . . — . . o. . g a 1 1 1 .1 . 11 . 1 * . . . 1 . . _ 1 . _ . . 1 _ 11 , . .. ... . . 1. .. I .I-.II.41.4I..I4II- -I14 I . . 1 1 1 .1. — ...1 _ .llIu 4 v Q. ‘l- O1|I|II' 41 ' i. .. .. w 4 41M ,1 +1, IQI’II 1l4-wo‘l- fl... I’l‘ll ’1? -‘lP-I’lLlll'z-I - .4 ...-III -.14-I4-1II-4-II .-..I..1- 1. . .1 1 . fl .1 a 1 1 1 1 .1 . ._ 1 IIITIIn III 1 . . : . . . .. 1. .1 1 . . _ . 1 . . _ . _ .. .1 . . 11 1 . _ _. . 1 . . 1 .x 1 t I . .I 1 1 1 , 1 . _. . . _ 1 l L . 1 4 p i‘l . (In {9+ a. 1 o 1 1 . 1 . _ . . _ IIIIIIl . . _ . . 1 . 1 . . . 1 I. n .1, .1. 13.11. , .. 1 1 1 1 1 .1wII1 Ir 1 1.. 1. . 1 1 1 . . _ 1 o .1. h -. Izlw. ..l..1~.01fil|1..vL.4oIOIIlIwIcII 1404441 III-MI 114¥l+l||112’4« III I .1.l4l v I. . I. II inII Ifl-.04I|IL.‘-IL.IIllnvchIIlM-IFI'UTIlftfli-Ill u, I. I - I. .. 1 .. H. .1 . 1 1 l 1 81 . _ . ,. 1111...... ,_ ._1 ...p. ‘rj' 1 -A 1 . 1 .1 . . .. 1 1 1 III IIIII .IIIrIIT-IIIIII .IIIIITIIII . _ 1 1. 1 . 1 1 ..1. IIIIIIIII A v. 91. O I” v 3‘ . n V ‘ d ~ — w u u h" h“. F I .. .- W "4 nI- ”A .IW . — fi - . n — F—/ “Ir “.1”. .N- r r d - _ 11h _ equilibrium line parallel to avoid a pinch in the column. The results were not very reproducible so several runs were made at each value of the amplitude and these were averaged. These data are tabulated in Table XXX, and presented graphically in Figure 2h. TABLE XXX EFFECT OF AMPLITUDE ON HETS AT FLOODING Packing Density 50.h lb/cu ft Packing Height lOl-in. CClu/HQO Flow Ratio 2.0 ’53 RPM 125 RPM Imfls Run No Amplitude (Inches) Average Run No Amplitude HETS Average 5 Runs 0 57.91 57.91 5 Runs 0 57.91 57.91 FthP 2 h1.6 ul.3 Fiu6P 2.0 13.8 13.8 Flh8P 2 u1.o F153P 5.5 5.7M 5.95 F152P 5.5 7.9 Fl53P(l) 5.5 6.16 F152P(l) 5.5 7.9 8.u1 F15uP 5.5 8.3u FflGP 555 95 FlSSP 9.5 5-87 F155P(l) 9.5 5.87 F156P 9.5 5.56 F156P(1) 9.5 5.56 5.76 FlSlP 9.5 5.90 F15lP(l) 9.5 5.8a Added complications entered into the runs on the 3.32-inch column that were not noticeable in the small column. The greater height of packing in this column, with the added increase in efficiency due to pulsation, has resulted in too many theoretical stages. This has caused the acetone to be extracted from the organic phase to such an extent that the throughput rates have been decreased. (It will be recalled that flooding rates are dependent on acetone concentration). An added complication is that these runs contain an increase in efficiency due to decreased throughputs as well as an increase in 1231. m N. .3... «7.4.144. s 2‘ nv -115- - 115 - efficiency due to pulsation. A correction would have to be made for this decrease in throughput before Figures 23 and 2h could be compared directly. A.great many authors have plotted graphs that show the effect of amplitude on column efficiencies. For example, Chantry et al (6), using a 1.5-inch packed column, show an effect of amplitude very similar to that obtained in this investigation. Cohen and Beyer (7), using a one-inch pulsed sieve-plate column, also observed a decrease in HETS with an increase in amplitude. Pulse Freguency Further series of runs were made on the 2.127-inch column at a constant amplitude to 5 mm to determine the effect of frequency on HETS. Two different flow ratios were used in these series as further assurance that flow ratio would have no effect on HETS values. The packing density was 51.7 lb/cu ft. These data are given in Table XXXI. TABLE XXXI EFFECT OF FREQUENCY ON HETS AT FLOODING 2.127-Inch Column Packing Height, 30.7S-inches Packing Density, 51.7 lb/cu ft 5-mm amplitude CClu/HQO CClu Rate approx. Flow Ratio 2.1 HETS 110 lemin HETS Run No Frequency (Inches) Run No Frequency (Inches) -- o h1.u -- o h1.h F86P 65 26.1 F8lP 65 26.1 F87? 125 16.2 F81P(A) 125 17.0 mm? 215 60 Fmrun 25 59+ These data are presented graphically in Figure 25. 5n n ...f [\r,wr....._.HH\_vmu wa._.r..~ .WIMWHVIWLNPnWWWuEWW 1‘ v |.|;~-3 m uh «.Hjbhv ... 1 p .L .1 . ...-I . » ..IH...‘ iulul-.nl.~JJ.H.Vvl.. .kw.‘ x...|f||[ -w'\ -oov Vf ‘v "4" (...- m L—J‘ _ 118 - It is apparent from the graph that HETS decreases greatly with an increase in frequency. The column efficiency appears to be more dependent on frequency than on amplitude. Similar results have been reported by Sege and Woodfield (A), working on a 3-inch pulsed sieve-plate column. They concluded that amplitude and frequency effects should be combined in a term afn where n has a value between 1 and 2. This, of course, shows the added dependence of column efficiencies on frequency. .A more exact measure of the individual effects of amplitude and frequency might be found if these values were converted to pulsed volumes and plotted against HETS. This was done by multiplying the cross sectional area of the Piston by frequency by amplitude (in inches) to obtain the cubic inches per minute of liquid displaced by the pulse piston. These data are plotted in Figure 26 and show quite clearly that column efficiency can be increased more by increases in frequency than by comparable increases in amplitude. This would indicate that the best column operation may be at high frequency and low amplitude. Miscellaneous The large column was equipped with expanded end chambers such as those recommended by Blanding and Elgin, while the small column had none. It is difficult to assign any real value to the presence of such chambers except perhaps for experimental purpose. Expanded end sections could be quite necessary in a pulse column if an exceptionally high degree of agitation is imparted to the two liquids. Such agitation promotes the formation of exceedingly fine droplets of the two phases and these are sometimes difficult to separate. The end chambers can, therefore, act as settling chambers where the - 120 - superficial velocities of the two phases are slowed down a sufficient amount to allow for disengagement of the dispersed phase. Expanded end sections may also be helpful when the interface is regulated very close to one end of the packing. Here they would act as a safty measure to prevent entrainment in case of minor changes in flow rates. The same advantage would be even more noticeable if the column were operated at flooding, where two interfaces are present to cause entrainment. Except for those conditions stated above, the presence of expanded end sections seems to offer few advantages. Thes conclusions should be regarded with some caution, however, because special spray nozzles or spargers may alter the column operation enough to warrant their use. The amount of free area in the packing support used in this investigation appeared to have little effect either on the allowable throughput or on the column efficiency. Apparently 50% of free area, such as that used for the small column, is sufficient to eliminate any noticeable effects of restricted flow through the packing supports. Little effort has been made in this investigation to analyze the economics of pulse columns. Other authors have considered this aspect in greater detail. Chantry et al (6) have concluded that the power requirements, even for a column several feet in diameter, are small; more important engineering considerations are pump size and vibrational stresses at higher frequencies. Jealous and Johnson (32) have derived an equation for calculating power requirements for pulsation. They have also recommended that a flywheel be used on the pulsator because the power is negative over half of the cycle. ther investigators have -121... pump for one of the entering phases. Comparison of the Two Columns Unpulsed Packed Columns The purpose of this investigation has been to determine what factors influence design. Each of these factors has been presented and discussed individually, but a comparison of the two columns of different diameters has been reserved for the conclusion of the discussion section, after most of the data have been presented. Little need be said here to inter- pret the influence of various factors on design because these have already been discussed in detail. Instead, the various factors will be utilized to answer the question of what might be expected in the design of large columns. An extrapolation was first made using Figure 22 at two different hypothetical packing densities to obtain HETS values for both the 2.127-inch and the 3.32—inch columns. These values, which have been corrected for end effects, are entered in Table XXXII below. Figure 10 was then used to obtain the expected throughput rates at these hypothetical packing densities. The values obtained for the 3.32-inch column were converted to a percent of throughput based on the 2.127-inch column. It was originally thought that Figure 17 might be used to get the expected HETS improvement due to this decreased throughput. Unfortunately the change in efficiency with reduction in throughput rates is not linear for unpulsed columns. Furthermore, the large diameter column also had a greater packing height and, since the total amount of mass transfer was greater in this column, the throughput rates were probably decreased because of mass-transfer effect even though the column was , ' _ . o \ , . . 7 . - 122 - operated at flooding. Just before flooding, more holdup of the discontinuous phases is expected; this in turn causes increased turbulence, with the result that higher efficiencies are obtained in the column. It is obvious from these remarks that it would be unwise to compare reduced throughput rates at flooding with reduced throughput rates when the column is not flooded. Since the two columns cannot be compared directly, linear approximations may be assumed for the comparison. For example, at a packing density of #7.5 1b/cu ft the larger column has a stage efficiency showing 18.5% improvement over the small column, but this is accompanied by a 12.5% reduction in throughput. .At a packing density of 50 lb/cu ft the efficiency is improved 12.5% and the throughput is reduced 15.5%. TABLE )C‘XII EFFECT OF COLUMN DIAMETER ON HETS VALUES Column Packing Sum of Diameter Density 1/2 % of Maximum (Inches) 51b 1 cu ft) KEEPS (ft/hr) Throughput 2.127 1+7. 5 81.1 18.l+6 . 100.0 3.32 1+7.5 66.1 16.19 87.6 2.127 50.0 72.9 17.16 100.0 3.32 50.0 63.7 1h.82 8.95 It can also be pointed out that increasing the packing density from h7.5 to 50 1b/cu ft for the small column gave NETS improvement of 10%, was accompanied by a 5% decrease in throughput. For the same increase in packing density in the large column, the HETS improved 3.6% and the throughput decreased 8.5%. 21.8 ‘1 I A»; ‘u. .o P- _ 123 - The ratio of tower diameter to packing diameter is 6.75 for the small column and 10.5 for the large column. From the above data it is obvious that increases in packing density give increases in column efficiency, but these are accompanied by decreases in allowable throughput rates. At a ratio of tower diameter to packing diameter of 6.75, the increase in efficiency is twice as much as the decrease in throughput. On the other hand, for a ratio of tower diameter to packing diameter of 10.5, the increase in efficiency is less than half as much as the decrease in throughputs. Although these are rather meager data, it may be concluded that channeling, if it exists, is more pronounced in the small column than it is in the large column. It can also be concluded that settling of the packing is advantageous if the ratio of tower diameter to Packing diameter is less than approximately 7, although it might be deleterious to settle the packing if this ratio is 10 or more. Pulsed Columns A.table similar to Table XXXII was made up to determine if channeling and scale-up factors could be observed for pulsed columns. This problem is complicated by the extremely high efficiencies due to pulsation as well as by other factors such as reduced throughput. Both columns contained so many stages that end effects were neglected. The only amplitude considered was 5.0 mm and the only frequency was 125 RPMS. HETS values were taken from Figure 18 which correspond to certain superficial velocities. These are tabulated in Table XXXIII. A value of HETS was taken from Table XX and corrected to 5.0-mm amplitude. These data are also added to Table XXXIII. - 12h - TABLE XXXIII COMPARISON OF HETS VALUES (PULSED) S-mm amplitude 125 RPM 2.1 CClg/H2O Flow Ratio Column Diameter Superficial % of (Inches) Velocity Total HETS 2.127 163.7 100.0 16.23 2.127 78.9 h8.2 10.10 2.127 21.6 13.2 3.70 3.32 no.0 28.1 7.0 These data are presented graphically in Figure 27 and show that pulse columns exhibit little noticeable channeling as might be anticipated. Comparison with Other Columns A review of the tabulation on pulse columns presented in the introduction of this report shows that, in general, pulse columns will give a minimum HETS of about six inches. That value also seems to be about the best that could be obtained in this investigation. Lower values were found, but only when the through- put rates were greatly diminished. If a value of 6 inches for a theoretical stage could always be anticipated for pulse columns, this would make an excellent design tool. 511343.: “Satanic: S... I}... - lehA - um“. xrsyser-4moeeem. : 010.6 .. ’1 "‘1 1 ...-1 £1 ,0 ‘”13 "'“W O .1 V i? 53 - PF 151-- a ‘1 «.- 2 1 a. 1 _ 1 1- 111-w 1 .1 .- .1. 141-11. . 1 O 1 Q 1 1 I 1 1**“'" 1 11.11%111. 101. 11 r. u 01.111 1.1-1.1.. 1111.11.1-fi1111i 7 1 1 1 1 1 1 17 1 1‘ 1 . . 1 1 «4 1 1 l 1 14 1 ‘.'"‘1"‘7‘7“*1fl"“ ‘ 1 11 ‘ .11. 4; . .4... -4p..- ......+....._ 1. I 1 11.111.111.111...” 1 1 .‘T 1" 1 11.1, 1 :L z . "1.: 1 1 1 1 1 111; a 1 .1 1 1 . . 1- .. .1 .11.-..1- . . 1 11+}! ._11.11I-1..111v11.1-11l-.1 11-11.11.111L1111 1.1-1.-.--..-4.....-.- . .. . . 1.11..... .. 14.: .1 .1 .. . 1 , . 1 - - .11..- 1 1 1 .5 1 .. . . 1'.l1 7-11.1{LY11+I|1 -1 --....r1.-1 1 1 1 1 1 ‘ ...111___"_ .- 4.-..v 4‘.—— _V-v._q'A..—.4 .-'1 1 .‘ ; 1 -1 .31 .1 1o... -1 1.1.11- . 111- 1 +4..... 1 1 71 .11.: .' J“ 1 1 1 .. 71.11 11-71.114.41. ' I 1 1 1 7 ‘ 1 11111 1 1 1 1 V e . . f ' 111' 14.11.114.11. D 1 1 I D 1 I 1 . 1 —-WO>~~«§ AT— -1 4—1-- ’. 7“-.':: '; . 1 . '; :; :;':*... .1 .: .. . 1.. :. . . 1 . . . 4 . ' I 1 .... . .. ~ .... ... . . .... . .. .. .. . ”. ..... .. .. ... . . ... , .. ... .- . 1 -11-.. 1 .1."---1 1 .- 1-H141.1 ..11 . .1111111 . .41. 1 - 11.1.1.le . . . . . W1 .- 1 4 .m... ... ..Hmn“... CI .1 ... H. .... . . . ..1.. o..> ... .. .. v. .... 1 .1 - 1s 4. .. 1 . . .. _ . .1 . . .- X . . ......u- . .. .. .. ..4 . .. . . . . a a o . . .. .‘ ... . ob . . 1 1- 1111?- L ..-1. .m. ...1........... H... .. . w 1 t.1~,. ; i-1-..- _.:w .. . _ o . A .. . .H... M s ..A - . . c a .. F . 5 .- 11.” ... 1 1 .... ...1. 1 .11 .. .121 -. .1 .. .. ....--. .1 . . .. .. - - ...-1.1...- L1-.. - . L .- . ...1 ..- . -I..IA.I1.1'| | OI - O . “ .4’1'A H .- . _. _ .11...i.. 4 . .....1. _. .. . 1 J 71...... - :,.. . ;d. .. .1 NM.. . 1 . . 1 . 1. .. .... .. T w .- 4. 11.11.111.11- 11.... ....... 171.1» 1. 1- .1- .1w141..1r1 ..- - . ~ . .. w .. 1. L . 1 .... H. ..r. - 1 . a . . . . M 1 1 . 1 !-.11 1114111H11111 .. v 0 11.1111 1 1 . . 1 1 ‘1 '1 ‘. . C .1. -9 ---—1y- ”+4 —*-'-.r>—~'-—¢-4F~—o—O—— 0‘ _ .1 -4 ' . 1 1 1 1 1 -4111. 1. - - . . . . 1 . .. 1 . . .11... .®\ 1 1 4. .4 . 1 _ _ .1... . . . 1 .. 1 w . 1 , , .. -..1 -. .1 .- .... .1. 1 - .1 -1 11-1 -1111. 11111-111111. 11-11-11.11.111-«11.11.111.111! 11.- ..1 1111-11- . 1 . 1 .1 _. . 1 1. ..1 . 1. . 1 . 1 1 4 . w H1“. . . a. — . 4 . H _ . 1 . ._ . 1 1r . e 1r 1 1. .. 1.-- a . 1. .1.- --11--1.1. -- . 11.11.11.117..- -..1- 11--- 1-4 a- 1 . T. .1. L1. 1-1111 - .- . 1 . 111 1 1 1- 1-111-111.1111 M g . ~ .~.. 4 ' 1 .. u a . o . . n . v 4 . ~ . n _ .r w 1r11.1fi.1 1 . 1 . . . . . . 1 . _ . . . h _ 1 l 1 l . 1 1 1 A '—Y 1 1 I . O r — .. -. .11 11.11111 1+3.-. 1-1- L.— ...... ..4 —-1.—9~-- »—1ko- - »&--—~- s o— “W .111L 111' 11'. 1.--.-. . n 1111111 1. 11111. 1|llflll‘tl “ ‘1 . 1 . . 1 . . 1 . . _ . K . _ . . 41-01-01. .1 ..--.1. -..-..-.1411. --w-_.-.----1fi 11.1.1111: . - 0-41111111: ........ #1-.-.- 1 1 .1 . w L 111 1 1 L n 111. 11 11 . ...“ W M” .— .M. _ 1 .1 . 1- .11 -1 .1 1--.: - 1..- - 1.1 t t. ... 1 - .. 1L 1- . 1 1 1 1 wzewaow.o:- was mos masses use: as soiuzemxoo1 . 1 .. L . L t .. 11.. .1 . . 1. 1 m whom“ ..H - 125 - CONCLUSIONS The following conclusions have been made as a result of this investigation: 1. The amount and direction of mass transfer has a very significant effect on the maximum allowable throughput rates. This is shown by the following: a. For the small column, unpulsed, when the entering carbon tetrachloride contained 1% acetone the maximum throughput rates were approximately twice those obtained when no acetone was present in either phase or when acetone was present in both phases in approximately equilibrium amounts. At a flow ratio of approximately four volumes of water to one of carbon tetrachloride, in the large column, practically no mass transfer occurs in the lower part of the column. This causes the maximum allowable through- put rates to decrease to less than half those obtained in the small column at comparable flow ratios. The large column gives about 30% less permissible through- put than the small column at flow ratio greater than 2.1 volumes of carbon tetrachloride to one volume of water. When sufficient acetone is present in the column, i.e., when the flow or carbon tetrachloride is high, the application of pulse permits increased throughput rates due to the greater amount of mass transfer. 2. f - 120 - e. When the efficiency of column operation is increased by using higher frequency or longer pulse stroke, further increases in the maximum throughput rates are observed. At the same time, the flow of carbon tetrachloride, which contains acetone, must again be increased to allow for the greater mass transfer. Almost a quantitative reversal of this effect occurs at lower flow ratios. 14.114. ’J I} l The equations of Hoffing and Lockhart, Breckenfeld and Wilke, as well as others, could not be used to predict maximum through- put rates when a solute was present. This was due to the fact I Jul-II": "...—stun!!! gem-Afi‘ .1... that when a solute was present the physical properties of the two liquids could not be measured at the non-equilibrium conditions existing in the column. Pulsing tends to gradually increase the packing density of pulsed columns; therefore, by their nature, pulsed columns cannot operate at low packing densities. High packing densities tend to decrease the maximum allowable throug’put rates, but these decreases are accompanied by comparable increases in column efficiencies. Pulsation can give as much as a 20-fold improvement in column efficiency. This, however, is obtained only under extreme operating conditions. At reasonable throughput rates and with vigorous pulsing, a stage height of approximately six inches (lh-fold improvement) was obtained. A summary Of the published data on pulsed-packed columns and pulsed sieve- plate columns contained in the beginning of this report shows that other investigators have also obtained similar stage heights. H_' -F , -127- Both pulsed and unpulsed packed columns show an increase in efficiency with decreased throughput rates. Pulsed columns appear to be less dependent on total throughputs than unpulsed columns. Unpulsed packed columns may exhibit an initial sharp decrease in column efficiency down to about 85% of the maximum throughput rate and from there give a gradual improvement with further decreases in throughput rates. This latter observation should be regarded with some caution, however, due to the extreme change in operating conditions at the onset of flooding. Stable emulsions did not form in any of these runs. At high amplitude and frequency, fine dispersions did occur, but these settled immediately when the pulse was stopped. Finer dispersions resulted in lower permissible throughput rates and better column efficiencies. New packing at first tended to change (perhaps to orient) with continued pulsing, giving a progressive improvement in column efficiency. The limiting throughputs remained essentially constant during this period. ter an initial series of runs had been made, no further change could be noticed. The HETS concept proved to be better than the HTU concept, because HETS was independent of flow ratio and feed concentration, whereas HTU values exhibited a pronounced dependence on flow ratio. ‘27:;— T. ‘fzw- fin-9“ '2! (C‘J‘u'VZ ..‘S 9. 10. 12. - 128 - At low and moderate conditions of pulse, both amplitude and frequency gave comparable improvements in column efficiency as Observed from the pulsed volumes. When more vigorous pulse conditions were used, frequency increases appeared to produce more improvement than could be Obtained with increases in amplitude. Mass transfer at the simple inlet tubes used in this invest- igation was not particularly significant (equivalent to only 0.15 NTS). This is in contrast to results reported in the literature where most of the mass transfer apparently occured at the ends of the columns. Maximum allowable throughput rates were sometimes increased and sometimes decreased by the application of pulse. This was true at certain flow ratios, regardless of whether or not a solute was present. Inconclusive results were Obtained in this investigation on the effects of scale-up. Apparently, a change in diameter from 2.l27 to 3.32 inches is not of sufficient magnitude to give a good comparison. A ‘.,.,__,,,_.._'.2 l , .- . 1". 'rc‘n- -— [1W APPENDIX I LIME-GLASS RASCHIG RINGS Weight per ring, 0.503 gms Displacement per ring, 0.1979 ml Area per ring, h07.2 sq mm Bulk Density Void Fraction Area lb/cu ft ftB/ft3 ft2/ft3 at 0.7228 17h is 0.6976 190 52 0.672h 205 BIBLIOGRAPHY Morello, V. 8., and N. Poffenberger, Ind. Eng. Chem., M2, 102l (1950). van Dijck, w. J. D., U. 3. Patent 2,011,186 (1935). Sege, G., and F. w. Woodfield, Chem. Eng. Progress, 50, 396-k02 (195A). Sege, G., and F. W. Woodfield, Int. Nuc. Eng. Conf., Ann Arbor, 6 (195M). Wiegandt, H. F. and R. L. von Berg, Chem. Eng., 61, No. 7, 183—8 (195k). Chantry, W. A., R. L. von Berg, and R. F. Wiegandt, Ind. Eng. Chem., u7, 1153-1159 (1955). Cohen, R. M. and G. H. Beyer, Chem. Eng. Progress, 19, 279-286 (1953). Griffith, w. L., G. R. Jasny, and Tupper, USAEC de class. doc. Report, AECD-3uu0 (1952). 9. Belaga, M. w. and J. E. Bigelow, USAEC declass. doc., KT-l33 (1952). 10. Billerbeck, C. J., et al, Ind. & Eng. Chem., AB, 183 (1956). 11. Murch, D. P., Ind. & Eng. Chem., h5, 2616-2621 (1953). 12. Feick, G. and H. M. Anderson, Ind. & Eng. Chem., it, not (1952). 13. Schuler, D. M., M. S. Thesis, Michigan State University (l95h). 1n. Thornton, J. D., Chem. Eng. Frog. Sym., Ser. 50, No. 13, 39-52 (195k). 15. Blanding, F. H., and J. C. Elgin, Trans. Am. Inst. Chem. Eng., 38, 305 (l9u2). l6. Hoepner, K., Z. Unters. Nahrungs Genuss Ms, 3h, h53 (1917). 17. Harasco, M., Ind. Eng. Chem., 18, 701 (1926). 18. Bennett, H., and F. K. Donovan, z. Analyt. Chem., 71, 207 (1927). 19. Bryant, w. M. D., and D. M. Smith, J. Am. Chem. Soc., 57, 57—61 (1935). 20. 22. 23. 27. 28. 29. 32. - 131 - Kremser, 1., Nat. Pet. News, 22, No. 21, h2 (1930) Sounders and G. G. Brown, Ind. Eng. Chem., 28, 519 (1932). Seidell, A., Solubilities of Organic Compounds, 3, Vol. 2, D. Van Nostrand Co., Inc., lBl (l9hl). Encyclopedia of Chem. Tech., Vol. 6, The Interscience Encyclopedia, Inc., N Y 135 (1951). Ballard, J. H. and E. L. Piret, Ind. Eng. Chem., 82, 1088 (1950). Hoffing, E. H., and F. J. Lockhart, Chem. Eng. Pregrcss, 50, 98-103 (1958). Elgin, J. C., and F. M. Browning, Trans. Am. Inst. Chem. Eng., : l 32, 105 (1936). Breckenfeld, R. R., and C. R. Wilke, Chem. Eng. Progress, M6, 187 (1950). rim. Crawford, J. w. and C. R. Wilke, Chem. Eng. Progress, A7, B23-31 (1951). Garner, F. H., s. R. M. Ellis, and w. H. Granville, J. Appl. Chem. (London), 5, 105-9 (1955). Treybal, R. E., Ligpid Extraction, McGraw—Hill Book Co. Inc. New York, (1951)- Sherwood, T. K. and R. L. Pigford, Absorption and'Extrnction, 2, McGraw-Hill Book Co. 1no., 836 (1952). Jealous, A. C. and H. F. Johnson, Ind. Eng. Chem., 87, 1159-1166 (1955). 17....- M'ifiifi‘EfllMMTu(EjflfgfiujiflMliflfll‘uWES