Document jmnR4YD72j7ekEG3Z65m7ZyG5

TOWOLDMON0030630 WATER_PCB-00015099 The performance of transformers and capacitors is generally based on the power factor. This quality is dependent upon the removal of adsorbable impurities. For many years Monsanto Company factories, as well as many customers have refined askarcl fluids by the simple agitation of a given quantity of Fuller's Earth with the fluid to remove adsorhahlc matter. The treatment was followed by filtration to separate the adsorbent from the refined fluid. When an adsorbent and fluid con taining adsorbable matter reach equilibrium, the equilibrium gen erally conforms to the Freundlich isotherm * = kO y where: x = the quantity of adsorbate, y " the quantity of adsorbent used, C = the concentration of the adsorbate remaining in the fluid at equilibrium, and k, "* -- constants characteristic for a specific system. The isotherm indicates that on one single equilibrium contact, a high degree of purification cannot be achieved. The equation indi cates further that with a sufficient number of repetitive contacts, the value of C may be made to approach zero (absolute freedom from adsorbable material). Since it is not practical nor possible to carry out sufficient repetitive batch refinements to achieve the maximum quality, one must resort to the next most practical approach; that being a packed bed which achieves these repetitive treatments on a single pass. To accomplish this, it is necessary to select the proper adsorbent and design a system with enough zones, which we shall refer to as equivalent transfer units (KTU), for maximum purifiea. tion. The sketches show a visual comparison between the batch and the packed bed system: As shown in Figure 1, the batch system reaches only one equi librium, that being dependent on the amount of adsorhahlc impuri ties in the fluid and the capacity of the adsorbent. The column system, Figure 2, shows the numerous equilibriums reached as the fluid passes through the column. When material passes through successive zones, the adsorbate collects in the bed. In each zone the consistently cleaner askarel contacts fresh adsorbent. In this way maximum purity is obtained. ObObBii TOWOLDMON0030631 WATER_PCB-00015100 Adsorbent Selection and Activation Although Fuller's Earth (Atlapulgus clay) has been used for years as the selected adsorbent, it does not function satis* factorily in the column system because of its softness. The gran ular material crushes and forms a mud when wetted with the fluid. The mud formation in a packed bed system restricts the flow to the point where the system is inoperative. To overcome the problems presented by tbe atlapulgus clay and to achieve other desired features, Porocel*, after extensive testing, was selected as the proper adsorbent. Porocel, being an activated bauxite, is considerably harder than the attapulgus clay which prevents crushing and even the fine material doesn't form a mud. Additionally, Porocel hits about 36% more surface area for effective adsorption than docs the clay. Best perform ance was achieved with 20/60 mesh although 30/60 or even 60/90 mcsli material will probably give satisfactory perform ance. The --60 or 200-|- mesh is definitely not recommended. At the other extreme, coarser than 20 mesh is not recommended. Before any adsorbent can be used for refinement, it must be thoroughly activated. The "degree of activation" is best deter mined as water content. For Porocel an acceptable moisture level is less than one percent (via Karl Fisher determination). If the adsorbent is packaged and shipped under fairly moisture tight conditions such as polylined bags or 6tccl drums, very little drying is necessary. The adsorbent can be dried in one of several ways. First, the adsorbent can be heated in shallow drying pans or a rotary kiln to a temperature of 450F to about 900F at atmospheric pressure and then charged to the adsorber. Second, the adsorber can be charged, vacuum applied (5 milli meters of mercury) and the adsorber temperature driven to about 300F. A third method, and probably tbe most expensive, is to fill the adsorber, heat to about 300F, and purge with dry or dry air until the moisture is driven oft. Since most plants handling askarel fluids have steam and vacuum facilities, the combination of heat and vacuum is probably the most econom ical way to dry the adsorbent. After the drying operation, the heat must always be cut back depending on the askarel being refined and the recommended temperature for refinement. Reg. trademark of Minerals & Chemicals Philipp Corporation. Adsorption Theory Each adsorbent granule, after activation, consists of a myriad of submiscroscopic pores running throughout the granule. Along the walls of these pores lies the active adsorptive surface. As an example of the great surface area and the 6mallness of the pores, one pound of Porocel con tains more than 24 acres of surface. The same equivalent for Fuller's Earth is 13 acres. The fitness of an adsorbent for a specific application is determined by the pore size, pole size distribution, pore volume, and surface energy. The volume of a particle is made up of the volume of the pores (1,,) plus the volume of the solid forming the balance of the granule. The density of the solid material excluding the pore volume is known as the crystalline den sity (I>c). Including both solid and pore volume is called particle density (Pa). The density of a particle is the reciprocal of its volume. In any column system you are concerned with a large mass of individual particles of irregular shape. The par ticles usually vary in size and size distribution between specified mesh limits. Due to the irregular shape, it is extremely difficult to obtain a three dimensional measure ment of these particles. Therefore, most mathematical explanations of packed bed columns are based on a shape factor (Qg) and a mean equivalent spherical diameter (P,,). This concept has allowed mathematical forecasting of many results with engineering accuracy. It is the equiv alent to treating the mass as if it were composed only of spherical particles of an average size. The mean equivalent spherical diameter is usually computed from the screen analysis of the adsorbent. In a packed bed, all the space between the granules is not filled. These intergranular spaces are called voids (E) and must be differentiated from the pores within the indi vidual granules. It is through the voids that the flow passes and it is through the pores that the mass achieves its adsorptive capacity, The Led density (Bu) includes the space occupied by the voids and that occupied by the mass of the particles. By applying vibrational energy during the packing of a fixed hed, about 15% more adsorbent may be placed in the same volume than by simply dumping the adsorbent. This can result only from a decrease in the void space with the consequent increase in the bed's effective density (Bx>). Such compaction of the bed must affect the course of adsorption since the void space forms the channels through which the fluid flows. The closer packing of the granules tends to reduce the transfer surface at the solidfluid interface, and at the same time, increases the unifonnity of the void channels. For mathematically defining the flowr, the average size of the void canals is assumed to be equal to the mean effective particle diameter. When the fluid containing adsorbablc constituents is passed through the bed, the flow is distributed in many different directions from that which it is introduced. Therefore, a given portion of the fluid may travel an extremely tortuous path and be subjected to continuous remixing with other portions of the fluid. The relatively rate of flow of such a portion of the fluid as compared to any other portion of the fluid is determined by the uni formity of the void canals. The canals arc most uniform when the hed is packed to its optimum density. Adsorption is a diffusion mechanism which docs not occur instantly. The adsorbate must diffuse first through tlie body of the fluid to the interfneial film between the solid adsorbent and the fluid. It must then diffuse through the interfacial film and finally by capillary diffusion reach the active adsorptive surfaces on the interior pores. Of the three, the capillary diffusion is the slowest and, hence, controls the rate of adsorption. Since the distance to be traveled by capillary diffusion is determined by the di mensions of the adsorbent particle, final rates must con sider the particle size and the exterior area of the particle available for mass transfer. With respect to the fluid, the rate of diffusion (Dj,) is governed by adhesion tension, interfacial tension, and the viscosity of the fluid (p). Only the viscosity can be evaluated accurately, therefore the quantitative treatment of adsorptive rates must he some what empirical. We have considered only what aflects the rate of adsorption. The capacity of the adsorbent bed is depen dent upon the pore size distribution and pore volume. The degree of purification is a function of the number of equilibrium contacts (ETU), i.e., the height of the bed (H). With these factors in mind we are now prepared to consider some practical evaluation of these principles. I _ _0^068^7 -II TOWOLDMON0030633 WATER_PCB-00015102 Operational Temperature Ak staled in adsorption theory, "invariably an increase in lcnij>crature lends to a decrease in the amount adsorbed, and vice versa",0' Ibc temperature becomes an important operating variable. From actual experience, the optimum temperature for maximum refinement bas been /ound to vary depending, on the fluid. Usually the operating tem perature is a function of the lowest temperature at which a reasonable fluid viscosity (p) can be maintained. As one may expect, because of poor diffusion (Dp), if tbe fluid is too viscous, tbe degree of refinement will actually drop. Therefore, temperature control in adsorptive refinement becomes mandatory. From actual experience we have found the following temperatures to be most suitable for refining the specific fluid. Fluid Aroclor 1260 Aroclor 1254 Aroclor 1248 Aroclor 1242 Mix of Aroclor 1260 1TCB,TCB, and dicyclodiepoxy carboxylate Trichlorobenzene Temperature (C) 135 116 85 70 30 25 0) "Fundmix-ntal Principles of Physical Chemistry," Pratton end Muien. r DESIGN CALCULATIONS Nomenclature for Symbols Used in Calculations Actual bulk density lbs/ft8 based on the bed E = Voids Fraction dimensionless Pd = Particle Density lbs/ft8 Dc = Crystalline Density Iba/ft3 Ip =2 Internal Porosity Dp rr Bulk Density % lbs/ft3 HETU = Height Equivalent ft Transfer Unit ETU = Equivalent Transfer dimensionless Unit JD Chilton -- Colburn j dimensionless -- factor for mass transfer JB := Chilton -- Colburn j dimensionless -- factor for heat transfer Re Reynolds Number dimensionless S* = Schmidt Number dimensionless k, = Mass transfer coeffi lbs/hr ft2 cient in a binary system G0 -- Mass Velocity of the lbs/hr/ft2 fluid calculated on the cross section of tbe empty vessel F = ^ where F -- volumetric ftVhr & flow rate of the liquid, and S = the cross sectional of ft2 adsorber a -- Surface area ft2 (of transfer area) ft8 (of bed) Dl = Self Diffusion Rates ft2/hr p r= Viscosity Ibs/fl.hr. g = Neutonian Constant p = Density of the fluid AP rr Pressure drop H = Height fni = Friction factor, a function of N1{ , 32.17 lbs (ft)/ lbs (sec)2 lbs/ft8 lbs/in2 ft dimensionless (modified Reynolds Number) Fig. 5-64, Perry's Handbook, page 5-51 Dp = Average particle diameter defined as the diameter of a sphere of the same volume as the particle N = Exponent, a function of the modified Reynold's Number NItt, Fig. 6-64, Perry*6 dimensionless Handbook, page 5-51 0 -- Shape factor of the solid defined as the quotient of the area of a sphere equivalent to the value of the particle divided by the actual surface of the particle dimensionless TOWOLDMON0030634 WATER_PCB-00015103 Design Calculations Equations HETU =- ^ VS / kxa ) = fa (where C,,= VS) G0 can be determined by knowing the fluid properties and the size on the adsorber. kx can be calculated from "Transport Phenomena" Jji ~ 0.91 IV06,Q. J,, ~ 0.91 Ro'061 Qs (Ro <50) Ji* =- Jn (by definition) G( cV where c =r total molar concentration V rr volume . . __ kv (Sc)2/S Jl>" Gc (Subslituling Ji, r J,,)-k*(',2A' =0.91 R,.~ 51 0, = h~ 0.9] Rc- 6 Q,G,, (S,)-2/" a is calculated from physical data on the adsorbents 6 (1-E) "/ Q.i>,. E is calculated from the following I*d Dc(Mp) Pfl(l-E) = Da I)c(Mfl) (1-E) Dn ,, Now Gc> kx, and (a) can be calculated. Therefore the HETU can he calculated. Since the total number of ETU is determined experimen tally (depending on tbc degree of purification desired), tbe total height can be figured. 11= HETU x ETU Typical Scale Up Example General Data Aroclor Viscosity (/t) of Aroclor 1242 (a) 70C -- 6.08 cp or 14.4 Ibs/fl hr Density (p) of Aroclor 1242 (ft; 70C = 1.35 g/cc or 84.5 lhs/ft8 Porocel 1,, (internal porosity) Du (bulk density) 0.50 (50%) 56 lbs/ft" D, (crystalline density) 206 lbs/ft Q 0.50 Dj, 20/60 mesh Mean Equivalent Particle Diameter 0.0019 fl 30/80 mesh Mean Equivalent Particle Diameter 0.0014 ft General Newtonian Constant g n 32,2 lbs (fi t /lbs (sec)2 Diffusion Rate (DL) = 3.1 X 10'6 ft2/hr Laboratory Conditions for Refinement Treatment temperature C = 70 Bed Packing 20/60 mesh Porocel Dp(ft) = 0.0019 Column diameter (inches) = 1.62 Column Height (inches) = 60 Bed Weight (grams) = 1850 Bed Volume (cc) -- 2050 Bed Density (Bz>) gm/cc =. 0.90 Flow Rate (cc/min) = 30 Go = Fp/S G0 = 380 lbs/hr ft2 R* = (Reynold's Number) Re - 0.053 Dl = 8 X ID 7 cmVsec 3.1 X 106 ft2/hr Sc = Schmidt Number PDL Sc = 5.5 X 104 kx = 0.91 Re'0'61 Q*G0 (Sc)'-78 kx = 0.485 ___ __Pfl -- p, = imi-u Pa= 206 (1-.5) = 103 ,, 103 - 56 E~ 103 E = 0.456 _ 6(1-E) Dp 0b a = 3436 ft2 HETU = ,G" kx a 380 " 0.485 (3436) i= 0.228 ft Since tbe total height of tbe laboratory system was 60 inches or 5 feet then, tbe total number of transfer units (ETU) is ETLU HKTU __ _ 5.0 " 0.228 =r 22.0 Since the laboratory system contained 22 units end refined satisfactorily, any scaled up system must contain this many units or preferably more. Assume we have a column available with the fol lowing physical dimensions that we desire to use. Diameter of tbe Column (ft' = 2.25 Height (II) ft = 12.5 Area (S) ft2 Volume (V) ft* =*(2.25)74-or4 = (4 ft2) (12.5ft.) ^50 Aroclor Data Treatment Temperature (C) Viscosity (ju.)at 7or{lbs/ft hr) -- 70 = 14.4 0406859 TOWOLDMON0030635 WATER_PCB-00015104 Density Bt70c (lbs /fts) (P) -- 84.5 Porocel Data Particle. Diameter D,, (ft) 0.0019 Red Charge (lbs ) 2800 Bed Weight 2800 lbs ... ,.8. 111) : :WVoluTMo="50f.-(lbs/fl)56 Absuihc we desire to refine 4.5 gpm of fluid Flow(F) = 4.5 gpm X = 36,lS/hr G,, =- F>/S = 760 ll.s/hr/fl2 r[=-g^ p = 0.100 Di= 3.) X 10 ftVhr xSc=-_p/*r>i. = 5.5 10* F,=- 0.91 R,r01 Q.G,,(SC)"2/8 k = 0.725 p __ _P<j -- Dp p 1>^ I),(l-Ir) ]*= 200 (1-.5) n =r 103 103 - 56 E = - 103 E = .456 _ 6(1-E) " _ D,, Q, _ 60-.456) * - 0.0019 (0.5) a = 3435 fi5 min = -kG - . _ 760 " 0.725 (3435) = 0.326 ft ETU & H HE'J'U = 12.5 0/0.326 (1 -- 38.3 units Using the existing 12.5 ft. bed column with a fluid flow of 4.5 g.p.m, would do a very adequate refining job, since only 22 units arc required to give adequate refining. The actual 38 units would probably give a slightly better quality. However the advantage in an excessive number of KTU's is the gain in operating time between shut downs to change out the bed and the turn up rale available on the column feed. I/Ct's assume wc want to double the feed on this unit to 9 gpm. The adsorbent properties remain tbe same. The only change is C0 which changes Rt,=__ G0I),, kx 0.91 Q.G,,(S,)-2/8 Then the HETU would change 11E1U ~ _G, k*a G0 = Fp/S G=(97X48)&15/4 G0 = 1520 lbs/hr/ft2 R0 = G0D,,/p 1520 (0.0019) _ R0=' 14 4 ~ U,^U Kx = 0.91 Rc-01 QsG0(Sc)-2/3 K* = 1.06 HETU = Go_ Kxa = 0.41 EUT - H HETU = 31 The answer is yes, the feed can be doubled. The total ETU's (31) is well above the laboratory value of (22) for adequate refinement. One can see from the equations that the adsorptive characteristics of the column are not dependent on the flow. However, by the same reasoning the pressure drop is shown to be highly dependent on the flow rate. Therefore pressure drop becomes o controlling factor in the design of these units. Red height, the adsorb ent, packing, and fluid temperature are likewise impor tant factors in the bed operation. Calculation of Pressure Drop Assuming the 12.5 ft. bed, packed with Porocel a! a bulk density of 56 lbs/fl3 and the fluid flowing at 4.5 gpm. According to Leva's equation for predicting pressure drop (AP) p 2 fn, G02 L(l-E)8-* ^ ~~ 144 Dj, p g Q88lN (E)8 _ where fw = the modified friction factor as determined by the Reynold's Number Rc L =- height of the bed (ft ) same as H g =r Newtonian Constant 32.17 lbs (ft) /Lbs force (sec.)2 N = exponent function of the modified Reynold's Number Rt. G0 = Superficial Mass Velocity in lbs /sec (ft )2 The remainder of the terms are explained in tbe calcu lation of HETU. 2(1000) (125) (l- 0.455)'-> MI ;00(>W> (84.5) (32.17) (0.5)M (0.45S) AP = 18.5 psig 0406860 TOWOLDMON0030636 WATER_PCB-00015105 Equipment The equipment necessary for an operational packed bed Byslctn consists of: 1. Feed System -- a temperature control is needed on the feed for the adsorber. A positive displacement pump to yield sufficient flow at the operating pressure is neces sary. 2. Adsorber--The actual adsorber shell must be a suit able material of construction such as aluminum or stain less steel to eliminate the possibility of product contam ination. The surface of a stainless adsorber should be cleaned free of any rust. The top of the adsorber should be removable for ease of loading the adsorber. The bot tom should be equipped with an adequate bed support to retain the granular adsorbent and yet allow the liquid to pass freely. The bottom should conlain a dump port large enough to allow for removal of the bed material. The unit should he adequately traced for temperature control. A temperature indicator should be installed in the adsorp tion zone. Vibrators should be installed on the shell of the adsorber large enough to pack the bed to optimum den sity. A N2 purge should be connected to the bottom of the adsorber and sufficient beat provided to drive off moisture in case the water content of the adsorbent is too high when the column is loaded. The normal recommended height to diameter is about 4-5 to 1 although a shorter unit could possibly be used. Less than 3 H/D is not rec ommended. The norma) safety items for a pressurized vessel are also necessary, 3. Filter--Small in-line polishing filters arc needed on the column to catch any residual adsorbent, etc., that may pass through the bed support. The filters should be on a by-pass arrangement to keep the unit on stream at all times. 4. Flow Controller -- A flow controller should be placed on the effluent of the adsorber to control the fluid flow tlirough tire adsorbent. 5. Storage System -- A storage tank for the finished prod uct should be provided with sufficient capacity to keep the column on stream. O*0*>#bl TOWOLDMON0030637 WATER_PCB-00015106 Adsorber Operation The column, after being throughly cleaned, is filled with activated (dry) Porocel. The bed is settled to an optimum density through the use of vibrators. The bed is then flooded with the fluid and allowed to soak into the bed and displace any air which may be trapped in the adsorbent. After the bed has liecn allowed to soak, pref erably over night at the operating temperature the for ward flow should then be started. Once the quality reaches an optimum level, the bed will continue to operate at this level until it becomes saturated with adsorbate. When the bed becomes saturated, the quality will show a sharp de crease or breakthrough. At this point the quality will drop to a considerably lower level and hold for a period of time and then drop to a completely unacceptable quality level. Ojicc the breakthrough is obtained, the bed should be changed and the cycle started over. f t This production model produces the highest quality refined askarel under the most economical condi tions. Because of the efficiency, these units operate continuously for months on a single bed charge, Shown above is a small laboratory model capable of producing up to 50 cc/min. of refined askarel. This unit can be used to produce laboratory quantities or collect scale-up data to design any size unit up to or larger than the unit shown below. Operating Costs Savings arc realized from 1. Manpower reduction 2. More efficient adsorbent utilization 3. Continuous faster refinement 4. liCss contamination 5. Less product loss C. Keduced maintenance The only disadvantage of the porocel column is the higher cost of the adsorbent (about 50%), but this is practically offset by the increased efficiency of the Porocel (40%). TOWOLDMON0030638 WATER_PCB-00015107 r~ Actual Field Experience The Monsanto laboratory tested a small unit collecting scale-up data for a pilot design. After a successful pilot demonstration, 8 pro duction unit was brought on stream yielding continuous refinement of the Askarcl (except for short intervals for changing the bed). Due to the efficiency of the system and the high quality product produced, Monsanto is adopting the system as a production standard world wide. In addition, the system is being disclosed to those customers desiring to use the most efficient and economical method of refine ment. Typical Results Achieved 1. Fluid Aroclor 1260 Aroclor 1254 Aroclor 1248 Aroclor 1242 Mixture of Aroclor 1260, TTCB, TCB, and dicyciodiepoxy carboxylate Trlchtorobenzene Resistivity (ohm cm)* x 10* 230,000 130,000 40,000 28.000 80,000 18,000 Conventional concentric electrodes, 0.1 Inch spacing 500 VDC stress with 1 minute electrification, @ 100" C. Hcsidcs the increase in quality, the column adsorber system was found to be 18 to 20 times mole efficient than old contact system. The increased efficiency in earth utilization of the column system reduces manpower requirements when compared with the old batch system. 1. Results are reported as resistivity which is the. reciprocal of power factor. At these quality levels, unless the new guarded electrode power factor system U used, an accuiate measurement cannot be made. COLUMN ADSORBENT EFFICIENCY# BATCH SYSTEM | COLUMN SYSTEM Compared to batch treatments -- assuming the same quantity of adsorbent (H06663 ! Jl TOWOLDMON0030639 WATER_PCB-00015108 A Monsanto Power Factor System constantly monitors on-stream quality through 100 cycle power factor at 100` C. This system requires no external sampling for quality control. Optional Equipment The Monsanto "in line" analyzer is a very useful instrument for monitoring the product quality from the column. The unit was de signed specifically for quality control on the effluent from this system. It provides a continuous monitoring of power factor and has provi sions for recording if necessary. The quality of the refined material is known constantly within three minutes after refinement. The "in line" analyzer is invaluable during start up of the adsorber to deter mine when recycle should be stopped and the forward flow diverted lo the storage system. The Information contained In this bulletin is, to our best knowledge, true and accurate, but all recommendations or suggestions are made without guarantee, since the conditions of use are beyond our control. The Monsanto Company disclaims any liability incurred in connection with the use of these data or suggestions. Furthermore nothing con tained herein shall be construed as a recommendation to use any prod' uct in conflict with existing patents covering any material or its use. Monsanto 0406864 TOWOLDMON0030640 WATER_PCB-00015109 Monsanto 0<i0b8fe5 1 Monsanto Company / Functional Fluids / 800 North Lindbergh Blvd. / St. Louis, Missouri 63166 1068-IGI-LRA LIT HO IN USA I TOWOLDMON0030641 WATER_PCB-00015110