You are on page 1of 92
UAE Liquid Agitation Reprinted with special permission from CHEMICAL ENGINEERING Cet Mca oe New York, N.Y, Defining the problem ts the first step in applying turbine agitators for mixing in which the continuous phase is a liquid. Developing an orderly procedure will enable the design engineer to specify an agitation system meeting the hydraulic, capital and cost requirements. How to select the optimum (7 An engineer today attempting to evaluate the re- quirements for turbine agitators has little information available for converting theory into sound practice. Our purpose in this series will be to present design informa- tion and a design procedure covering size selection, mechanical design and cost estimation for process ap- plications of turbine agitators. ‘These techniques will enable an engineer to make: () preliminary estimates of turbine-agitator size and Cost fora proposed new proces, (2) provide a basis for analyzing existing agitation equipment in a plant modernization study, or (3) provide a bai for ape. ing turbine agitation equipment, ‘The following listing of major copies uttnes he subject matter to be covered for the proper application OF turbine agitators pepeepe © Design procedure: organization and logic. ™ Fundamental concepts. © Design procedure applied to blending-and-motion, solids-suspension, and gas-dispersion problems. Mechanical design, § Economic evaluation and optimization. A turbine agitator is a mechanical device that pro- duces motion in a fluid through the rotary action of impellers. The turbine impeller consists of fixed-angle blades attached to a hub dfiven by the agitator shaft. ‘Turbine impellers are characterized by the type of flow produced, Axial-low turbines move the fluid par- 2 —GHRTCAL ERCINEERING DECENOER Ss Lewis E, Gates, Terry T. Henley 7 allel to the axis of the agitator shaft, while radial-low turbines discharge the fluid from the impeller region perpendicular to this axis, ‘The most common axialflow turbine, the 45° pitched-blade turbine, is shown together with its basic flow pattern in Fig. 1. Although there is a slight radial ‘component to the flow, this pattern is primarily axial flow. A typical radial-low turbine, a flat-blade impel- ler, is also shown in Fig. 1 along with its representative flow pattern. The fiat blades can be mounted on a hub, as shown in Fig. 1, or on a disk if flexibility in the number of blades or impeller diameter is desired. The design problem for turbine agitators Design of most types of process equipment generally involves defining and meeting a well-understood proc ess objective. For example, the design of a distillation column would begin with’ a statement of the desired degree of separation. Process conditions and number of theoretical trays would then be established. The equipment designer would calculate diameter and height of the fractionation column and the number of trays by using established correlations for hydraulic performance and tray efficiency. ead John Tene, Ohemineer, Tne Tn agitation equipment, a comparable procedure for specifying and designing has not been established in the published literature. ‘The reasons for this include the wide range of applications for which agitators are used, the lack of agreed-upon criteria for agitation performance, and the relative complexity of many agi tation applications in the chemical process industries (cep. Because of these difficulties, a specification as written for an agitator often includes such imprecise statements regarding the degree of agitation as: Provide sufficient agitation to promote reaction. Promote contact of solid and liquid Blend two liquids to uniformity. Disperse gas to promote reaction. Provide (mild, medium, violent, vigorous) agita- tion. With such vague criteria for performance, a precise selection of an agitator is not possible. ‘The function of the design procedure to be presented in this series is to provide a technique for improving communication of agitation requirements, as well as describing how this information can be converted into the proper hardware. In order to understand the various technologies re ‘CHEMICAT ERREERING DEEN Blade pitch establishes type of flow for turbines quired for the design of turbine agitators, we will ana- tyze the organization and logic presented in the flow chart of Fig. 2. The principal sections of this logic, as noted on the extreme left of Fig. 2, will serve as head- ings for the following discussion. Classification of the agitation problem Applications of turbine agitators in the CPI ‘one or more of the following objectives: Bulk mizing—Combining process liquids of dissim Jar composition and properties 8 Chenical eaction—Distributing reactants and prod: ucts to promote desired reactions ‘Heat trensfo—Increasing convective motion adja- cent to the transport surfaces 1 Mass tansfer—Promoting contact between separate phases and different compositions, 1 Phase interaction—Suspending solids or dispersing {gates and immiscible liquids. ‘Analysis of these process objectives indicates the re- quirement fr generating fluid motion to contact li uids, solids or gases in a liquid phase that is continu. ‘ous. The phases present in the liquid to be agitated permit the first major organization of the design proce: lure into the categories of blending and motion, solids suspension, and gas dispersion. Referring to Fig. 2, we can easly see that the blend- ing-and-motion design category should be used if only a 3 oN rhe re imegianon liquid phases are present, the solid-suspension category if only solids and liquids are present, and the gas dispersion category if only gases and liquids are in the system. Although combinations of these categories may bbe encountered in many processes, separate treatment of them will normally establish the most diffieult, and therefore contolling, problem in terms of equipment selection, Magnitude of the agitation problem “The magnitude of the agitation problem isa function cof how much material is to be agitated, how difficult itis to agitate, and the intensity or degree of agitation, required. Size and difically The mass of the phases present is fan important indicator of magnitude. Due to the CPL convention of stating agitation problems in terms of volume, an equivalent volume Vj., will be defined in the following equation as the product of specific grav- ity, S,, and actual volume, V, gal: Veg = SV @ is definition has the advantage of retaining the units of volume but also of being proportional to the ‘mass of the phates present “The variables that are used to define the degree of difficulty of agitation are: viscosity in blending and motion, solid-ettling velocity in solide suspension, and gas-flow velocity in gas dispersion. These terms will be referred to in the design procedure as primary variables. ‘The effect of an increase in either the equivalent volume ora primary variable is an increase in the magnitude of the agitation, problem. Required process reult—After defining the quantity of material to be agitated as well as the primary variable, an examination of the desired process resul is necessary A process result of fuid agitation can be described by a wide range of chemical engineering terms having quantitative meaning, such as heat-transfer rate, mass- transfor rate, blend time, degree of blending, reaction rate and yield. It would be very desirable to design the turbine agitator to directly produce this result, As pre- viously noted, it is often impossible to state the desired result with precision, However, the designer should attempt to define the desired process result to the extent that itis possible. Requivad dynamic response —IE the process result can not be defined, what can be said about the degree of agita tion desired? To specify the degree of agitation, we will adopt a new term called dynamic response. Dynamic response describes the resulting agitation in purely physical terms. Dynamic responses for blending and motion, solids suspension, and gas dispersion in procedure are bulk fluid velocity, level of solids suspen sion, and degree of gas dispersion, respectively. To illustrate this concept, a low level of dynamic response for blending and motion would be a very minimal velocity throughout the agitated fluid, whereas 2 high level of dynamic response would be very high velocities, Tn solids suspension, a ow level of dynamic response ‘would correspond to Solids just minimally in motion on the bottom of the vessel, while a high level of dy- namic response would be sol throughout the fluid, ‘A minimal level of dynamic response in gas disper- sion would be relatively large gas bubbles rising rapidly to the liquid surface and poorly distributed throughout the liquid phase. A high level of dynamic response would be gas bubbles finely dispersed and distributed throughout the process veseel As a convenient approach to design, the dynamic response has been related to a scale of agitation that varies from 1 to 10 for the majority of practical appli- cations for turbine agitators. This scale serves as a com munication tool to indicate the level of dynamic re- sponse desired, uniformly dispersed Equipment selection for dynamic response Up to this point in our procedure, we have empha sized the process requirements and the associated dy- namic response. The purpose of the procedure, however, is not fulfilled until industrial equipment that will meet the process requirements has been designed and se- lected. Agitator drive—Specification of a turbine agitator so that detailed mechanical design can take place requires definition of (1) prime-mover power, (2) rotational speed of turbine (ie., shaft speed), and (3) number, type, size and location of the turbines. Since rotation of the. impellers in the fluid produces the dynamic response, it would be logical to first define the impeller system and its rotation speed. The power required to acsomplish the rotation of the impellers ‘would then be determined, However, mechanical design fonstraints (including shaft design, standardization of prime movers and drives, and economic considerations) limit the number of combinations of power (for prime mover) and speed (for impeller rotation) that are available. It is more Convenient to prepare selection tables in which practical combinations of power and rotational speed are tabu lated asa single function of equivalent volume, primary variable and dynamic response. Inpeller system Design of the impeller system inside the process vessel requires specifying the type, number, location and size of the turbine or turbines. The appl cation establishes the impeller type. Pitched-blade tur- bines should be chosen for blending-and-motion and solids-suspension applications and flat-blade turbines for gas dispersion. Selection of the number and location of the turbines begins with a calculation of the Z/T ratio where Z is the liquid depth and Tis the tank diameter, as shown in Fig. 3, nd continues with finding the number of impellers and their position in the fluid in order to control that geometry. Specific design recommendations for the number and location of the impellers as a function of system geome try will be made in future articles of this series. Caleu tion of the impeller diameter, D, to fully use the eapa- bility of the prime mover and drive, is perhaps the most critical element of design for the impeller system. The theory and practice of the required calculations will be covered in future articles dealing with fundamentals and applications. ‘Shafts and seals After the system geometry has been —GERTCAL EROINTRTING BEER TE 5 Design begins with analysis of vessel and impeller geometties for the agitated system analyzed and the number and location of impellers has been fixed in the process fluid, the calculation of total shaft length, L, and turbine spacing, S, can be com- pleted. Proper mechanical design of an agitator shalt 's extremely important. Structurally, the shaft must be of sufficient size to transmit the rotational load, com- bined with bending loads generated by random hy- draulicimbalances, acting on the impellers. In addition, an analysis of the natural vibration frequencies of the shaft and impellers must be made in order to assure that these frequencies are sufficiently far from the ‘operating frequency. Operation of the shaft and im- peiler system at their natural vibration frequency ean ‘reate forces that are destructive to the equipment. The complete theory and application of that theory to shaft analysis will be coyered in articles discussing ‘mechanical design. The practical implication of the shaft analysis i that for a given power and shalt speed, there is a total shaft length that cannot be safely ex: casded. Because chemical process equipment commonly op: erates at clevated temperatures and pressures, one of the key clements in rotating equipment is seal design Future articles in this series dealing with mechanical design will amphasize shaft seals ranging from common lip seals and stufing boxes to complex mechanical seals Seal design will logically occur after shaft design be- cause of the variability in seal design with changes in shaft diameter Economic evaluation Engineered equipment is usually a careful balance of performance and cost. Therefore, any design proce. dure that does not provide for cost estimation would bbe of limited value, Extensive cost-estimating proce- dures for turbine agitators will be featured in later articles. Normally, alternative designs for turbine agitators are mechanically feasible, and any one of them will achieve the desired agitation. The concept of a single answer for the power required to solve an agitation problem is incorrect. In most applications, several different combinations of power, speed and impeller system can be used to give the same result. The choice then becomes an economic one in which eapital cost for equipment and operating costs are significant. In creasing cost and availability of energy must also be luded in the evaluation, The articles to follow in this series will allow (a) analysis of the required performance, (b) selection of the prime mover and drive to give that performance, and (c) complete mechanical design; cost and opti- mization for an agitator. The next article of this series will appear in the issue of Jan. 5, 1976, and will deal with velocity and dimensional analysis, flow patterns, power correlations and blending ‘The authors ss mn ot gpieatign engine fr the Agitator BEST eminent, BO Bee 1123, Dayion, OH S50, He ana BS ad se hts la ehemicl ening fom repre’ Prokaioal cgiee in Ohi, “fled master’s Gear progr in ‘echanical engineering atthe Scien of Bayon Sek ee ip Ine Hle shed vera far th the ie eh ise ces ea inkenity Hees ter FE Taree Ge'vorking in rasareh and icons eth 6 CHEN ENNIS BEER TE oO Dimensional analysis for fluid agitation systems Dimensional analysis converts a large number of geometrical, operational and physical variables into a small number of significant groups that form the basis for design methods. David 8. Dickey and John G. Fenic, Chemineer, Ine Ci The fundamental concepts of liquid motion and ‘other transport phenomena establish a framework for agitator design. The design procedure, outlined in Part I of this series (/] uses the results of fundamental analysis along with practical experience to determine agitator requirements. ‘Our intent is to create an engineering awareness of the concepts behind agitation before presenting details of the design procedure. The articles on fundamentals Are not intended to present an exhaustive survey of the literature on agitation since other sources are available [23], Instead, our emphasis will be on fundamental information that forms the basis for design methods to bbe described in future articles. Dimensional analysis Dimensional analysis reduces the number of inde- pendently variable quantities that describe a problem by combining the variables into dimensionless groups Although fluid agitation involves a large number of ‘geometrical, operation and physical variables, a rela- tively small number of physically significant, dimen- sionless groups can be established. ‘One method of dimensional analysis collects all of the seemingly relevant variables, and systematically arranges these variables into dimensionless groups. A better method, and the one used here, establishes mathematical model that describes the important phys- ical phenomena and then rewrites the resulting equa- tions in dimensionless form (f] Analysis of a model significantly reduces the chanées of overlooking an important variable, and also provides considerable insight into the physical significance of the dimensionless groups. The groups that we will discuss in this article are summarized in Table 1 Navier-Stokes equation Fluid motion in an agitated system must obey the laws of conservation of mass and momentum, These laws can be written as an equation of motion that describes the velocity and pressure distributions within the fluid For a constant-density, Newtonian liquid, the Navier-Stokes equation for a mass-and-momentum balance in terms of local presture and velocity (5) is Do Wt PDE = TEP + HU + Og “ RRTARYS TE 7 Cara Dimensionless variables that are a ratio of the actual variable to a characteristic quantity will be substituted into Eq. 1 Characteristic quantities are selected to represent the principal dimensions of length, time and mass. The characteristic length used in agitation is the impeller diameter, D. Characteristic time is the reciprocal of the agitator rotational speed, 1/N. Characteristic mast is the product of liquid density, , and the cube of the impeller diameter, D3, Characteristic velocity can be derived from the length dimension and time dimension by using the product of impeller diameter and agitator speed. Dimensionless lengths and dimensionless time are defined as: = x/D @ in ‘The differential operators can be made dimensionless by combinations of the characteristic length and time. Dimensionless velocity is the ratio of actual velocity to characteristic velocity: vt = o/ND @) Dimensionless pressure can be defined from the char- acteristic quantities for length, time and mass, and the gravitational conversion factor, g,! @ = Pose Ne? ’ @) where the reference pressure, fy, is selected to simplify boundary conditions on the model. Substituting these dimensionless variables into Eq. (2), and rearranging coefficients, yields a dimensionless form of the Navier-Stokes equation, which is descriptive of an agitated liquid: Dot _ pape H Dears [sé Be P+ Ls] + In Eq, (5), two dimensionless groups appear as pa- rameters. The Reynolds number for agitation, D2Np/, appears in reciprocal form as the coefficient for the viscous dissipation term. This Reynolds number repre- sents the ratio of inertial to viscous forces. The second ‘dimensionless parameter is the Froude number for agi- tation, DN?/¢. This represents the ratio of inertial to {gravitational forces. Analysis of Eq. (5} also indicates the basic relation ships for velocity and pressure. For a given set of initial and boundary conditions, which implies geomettc simi- larity, the velocity and pressure distributions ean be ex Ifo pressed as functions of the Reynolds and Froude numbers: 254.45") = SWae Nee) © PN 99) = S (Nae Np) a Where the liquid surface is essentially fat, such as in fully baffled tanks, gravitational effects can be elimi nated. Velocity and pressure distribution are then de- 8 ——SHEUCAL NONREERE ROS termined solely by the magnitude of the Reynolds number: v5 "6 0) = F(Nae) ® PM EN) = fps) ® These results show how four. sccmingly independent variables—diameter, rotational speed, density and vis- cosity—are combined in the Reynolds number to sim- plify the functions for dimensionless velocity and pres- sure distributions Other dimensionless relationships Mathematical relationships can be written for other transport phenomena in agitation. An energy balance describes the temperature distribution within a uid, assuming constant properties for density, heat capacity and thermal conductivity. When this relationship is written in dimensionless form, a new paraineter ap- pears: Dee [ 1 Det UN aN, ‘This parameter is the product of Reynolds and Prandtl numbers, and is known as the Peclet number for heat transfer (Table 1). The Prandtl number represents the ratio of momentum to thermal diffusivity and involves only liquid properties, Establishing initial and boundary conditions for both the equation of motion and. the energy balance, Eq, (8) and Eq, (10), and neglecting gravitational effects, yields a dimensionless temperature distribution that is 2 function of only the Reynolds and Prandt! numbers: “SNe Nor) a) Exq, (11) simplifies the parametric investigation of heat transfer in an agitated liquid. Mass transfer can be analyzed by writing a compo- nent mass balance for the individual chemical spedics. ‘The dimensionless form of the equation is analogous to Eq. (10), except that temperature is replaced by dimensionless concentration, x,*, and the Prandtl number is replaced by the Schmidt number (Fable 1, ‘The Schmidt number represents the ratio of momen- tum to molecular diffusivity Analysis of the mass-transfer equation establishes the dimensionless concentration distribution as a function of Reynolds and Schmidt numbers: £47519 "5 4%) = S(Nee Nae) (12) Eq, (8), (9), (11) and (12) describe the general behav- ior of velocity, pressure, temperature and concentration in an agitated vessel. In subsequent discussions dealing with power, velocity, blending, and heat and mass transfer, we will show how to apply these conclusions to practical. problems. Agitator power Pressure distribution throughout an agitated vesse! cannot be applied directly to design, but one portion of the pressure distribution along the face of an impeller blade can be related to the power requirements of the agitator. his relationship comes from a description of the Jeo (io) O38 41%) Nomenclature D/Ot Substantial time derivative (ie, derivative =o ; Dimensionless pressure in Eq. (4) Dimensionless pressure in Eq. (21) Power Pumping capacity Blend time Dimensionless time Dimensionless blend time Tank diameter imensionless temperature ocity Dimensionless velocity Dimensionless concentration Dimensions of position W Impeller blade width Viscosity & ensity Vector differential operator V2 Laplacian operater physical system, Power is the product of rotational speed and applied torque. Torque is determined by integrating the pressure distribution over the surface of a flat-blade turbine. Therefore, the relationship be- tween fluid pressure adjacent to the blade and power ( — Podsage © P/ND® a3) Substituting Eq, (13) into the definition of dimen- sionless pressure, Eq, (4), yields an important rela- Conship between dimensionless pressure and power: pre: (Ge) aie Poe iD’) awa? = GNADE The expression on the right-hand side of Bq, (14) is the power number (6). Substituting the power number into Eq. (7) shows that it must be a function of the Reynolds and Froude numbers: (14) Wee Ney} aed ‘The power number relates imposed forces to inet forces ‘When gravitational effects are not a factor, the power number of Eq. (14) may be substituted into Eq. (9): i pags = 1M) a) 2q,(16) is normally uted a basi forcoveatng data fo agitator power The limiting cases ‘The limiting cases for agitator-power requirements involve large and small values for the independent parameter, the Reynolds number. Large values of the Reynolds number indicate turbu- lent agitation where inertial forces dominate viscous forces. Neglecting terms in the Navier-Stokes equation, Eq, (1), for viscous and gravitational forces gives an expression for fluid motion known as Euler's equation [5]. This can be written in dimensionless form as: Dot _ pays pe = 0p a ‘Since the Reynolds number is no longer a parameter, the pressure and velocity distributions are fixed for this limiting case. A fixed, dimensionless pressure distribu- tion along the impelier blade means that the power number is a constant at high Reynolds numbers: Pg constant A airigs = connant (18) Rewriting Eq. (18), we find that Pee pN*D® 19) ‘This limiting case is typical cations of turbine agitators. ‘The other limiting case corresponds to small values of Reynolds numbers. A viscous or laminar form of the Navier-Stokes equation neglects terms for inertial and gravitational forces, and results in an equality between Pressure and viscous forces: aX p = pV 0) Analysis of Eq, (20) is done with a different form of dimensionles pressure, The characteristic pressure it related to viscous force per unit area, since momentum hhas been rieglected such that (= Polte f aN en Substituting the dimensionless pressure into Eq. (20) reduces it to n a majority of appli- o Deptt = Vetee (22) ‘The Reynolds number has again been eliminated as STS 9 PaO Povier number i vious ee | Reynolds eumber, N/a — \ “Transition a. Power numbor | | | ! | | \ 1 i { "Difnensionles velocity, 9/ND-— STREET IT Transition ‘Turbulent Reynelds number, 02M fb — Pumoing nimber, Q/ND3— > Reynotda number, D°No)i= |e Pumping eurmber |. A Reynolds number correlates dimensionless parameters for the Fig. analysis of turbine-agitator systems 10 2 parameter, and the dimensionless velocity and pres- sure distributions are constant. Substituting the rela~ tionship of Eq. (13) into Eq. (21) gives: ce (2) fe x Phe wre (ioe) = avaos a “The group on the right side of Bq. (28) ean be con- sideced a “viscous power number” (Table 1) Since it is a constant in the viscous range: Pee pN*D? en From the relation for power number, Eq. (16), and the limiting cases for high and low Reynolds numnber, the general form of the relationship between power number and Reynolds number can be plotted on loga- Fithmie coordinates, as shown in Fig, 1a. The general shape of this curve can be derived directly from basic fluid dynamics without reference to specific experi ‘mental data. Power correlation Extensive experimental investigations have shown that power number is a function of Reynolds number for geometrically similar systems [2], as predicted by dimensional analysis, Eq. (16). A single relationship between Reynolds number and power number corre: lates the effects of density, viscosity, rotational specd nd diameter. The power number correlation has been determined for numerous impeller geometries [2,5,7- 104, Since the power number was developed from an expression [such as Eq, (13)] for the pressure distribu- tion along the impeller blade, it is not surprising that the correlation depends primarily on the impelicr con- figuration. Correlations of experimental data for four turbines are shown in Fig. 2 [7}. The turbines vary in the method of mounting the blades to the shaft, the blade angle, and the ratio of blade width, M, to im- peller diameter, D. For high Reynolds numbers, the turbine with disk- mounted blades has a slightly higher power number than that with hub-mounted blades, as seen by the difference between Curves 1 and 2 in Fig. 2. A signifi cant effect on power number is observed for turbines having different width blades, Curves 2 and 3 in Fig 2. Finally, the pitch or angie of the turbine blades affects the correlation between power number and Rey- nnolds number, Curves 3 and 4, The vertical projected height is used as the blade width for the pitched-blade turbine, Other geometric variables also have an influence on power, They are generally less significant than the ge- fometry of the impeller itself. Such variables include impeller clearance from the bottom of the tank, ratio of impeller diameter to tank diameter, impeller spacing in multiple-impeller systems, and number and size of baffles. The effects of these variables are discussed else where (2) ‘The power correlations in Fig. 2 indicate three ranges of liquid motion: turbulent, transition and viscous Fully turbulent agitation occurs above a Reynolds number of approximately 10,000. The viscous or lami- nar range occurs below a Reynolds number of about ‘Turbine design affects power requi 20. The range between these limits can best be de scribed as transition flow since flow patterns change depending on the Reynolds numbers. In agitation, transition flow occurs over a broader range of Reynolds numbers than is typical of flow in pipes The correlation in Fig. 2 shows that the power num ber for any particular geometry becomes constant in baffied tanks. This confirms the limiting case discussed in dimensional analysis for high Reynolds number. The turbulent range is important for agitator design, since nearly all low- and medium-viscosity applications of ustrial equipment operate in this range. In the viscous range, the expected result of power number being inversely proportional to Reynolds num- ber is also confirmed by experimental data, For each of the turbines in Fig. 2, a logarithmic plot of the relation between power number and Reynolds number has a slope of —1 in the viscous range. Fluid velocity Understanding the behavior of fluid velocity in an agitated vessel is as important as understanding power, since by definition agitation is the fluid motion pro- duced by impeller rotation. This point is frequently misunderstood. The use of power, or power consumed per unit of volume, is often applied to characterize the degree of agitation. In this and subsequent articles, we will explain why the same degree of uid motion can be achieved with different power levels in the same vessel, and why less power per unit volume is required sments for agitated systems for equivalent motion in large vessels as compared 10 smnall ones. The limiting cases Behavior of velocity distribution for the limiting cases of turbulent and viscous flow can be established in a manner analogous to the development of these cases for power. Referring to simplified forins of the Nav Stokes equation for turbulent and viscous conditions, Eq, (17) and (22), we find that the dimensionless veloc. ity distribution is independent of Reynolds number. Thus, velocity distibution is constant at very low Reynolds numbers, and is also constant at very high Reynolds numbers, for a particular set of initial and boundary conditions. We can represent this conclusion graphically, as shown in Fig. tb. In the viscous range, dimensionless velocity, v/ND, has a smaller magnitude than it has in the turbulent range. This is physically reasonable because for a given value of the product of rotational speed and impeller diameter (ic, ND), it would be expected that actual velocity would be less in the vise cous range than in the turbulent range. Flow patterns in agitated vessels In fluid agitation, the direction as well as the magni- tude of the velocity is important. The directions of the velocity vectors throughout an agitated vessel are, by definition, the flow pattern Since the velocity distibi- tion is constant in the vitcous and turbulent ranges, the —SHRACAL TRGIRFRRING TINTARY Se io ‘4. Axia-flow pattern ‘with pitched Blade », Raga fow pattern vith fat blade Blade pitch determines flow pattern Fig.3 flow pattern in an agitated vessel is fixed, Flow patterns at high Reynolds numbers have particular practical importance “The flow pattern produced by a pitched-blade tur- bine in a typical industrial unit is shown in Fig. 3. Although there is a smail radial component to the discharge velocity from the turbine, the flow is primar- ily axial. This fow pattern is significant in solids sus- pension, where the dircetion of flow tends to sweep solids off the bottom of the vessel. As the Reynolds number is reduced into the transition range, the radial component of flow increases. At still lower Reynolds numbers (approaching the viscous-low region), the radial component is reduced and the flow pattern again becomes predominantly axial ‘The turbulent flow pattern of a typical flat-blade turbine is also shown in Fig. 3. The predominantly radial discharge of this turbine is particularly effective in gas-dispersion applications. Velocity and pumping Measurements of velocity have been summarized in the literature (2/7). In many of these investigations, velocities have been measured in order to develop cor. relations for pumping capacities of turbines The conclusions regarding velocity behavior in an ‘agitated vessel can be applied directly to. pumping capacity. If the relationship developed for velocity bee havior In Bq, (@) 8 applied to an average velocity, the result is: Paay/ND = f(Npe) (25) The relationship between average velocity across an area and pumping capacity through that area is: Oayg = Be 'a0 = B 6) Substituting this relationship into Eq, (25) yields @by)_ a _ ND 7 Np? =/(Nee) en The behavior of the pumping number, Q/ND’, would be expected to parallel velocity behavior, includ- ing the limiting cases of low and high Reynolds num- her, as represented in Fig. le These concepts of velocity and pumping will be ap- plied to blending-and-motion problemsin a later article of this series. The form of the dimensionless velocity and pumping-number curves in the low-Reynolds- number region is of little practical importance for tur- bine agitators, but the characteristics of the curves in the transition and turbulent regions provide the basis for blending-and-motion design. Blending and blend time mn a discussion of blending and blend time, a distine- tion must be made among the terms: blending, mixing agitation. Blending refers to the intermingling of imiscible fluids to produce some de¥ree of uniformity. Mixing involves the production of uniformity between materials that may oF may not be miscible. (For miscible liquids, mixing and blending are synonymous.) Agita- tion isthe most general term, involving the production ‘of uid motion for blending, mixing, heat transfer, mass wansfer, etc, “The process of blending on a microscopic scale is not Methods for determining blend time Table I! 12 —GHERIEAT ENGINEERING TARTAR well understood, although much fundamental work has been done. The chapter by Brodkey in Ref. 2 provides, excellent background on this subject. However, in agi- tated vessels, the developments in microscopic mixing hhave had limited application to, practical design be: cause most blending applications are controlled by bulk. mixing rather than by microscopic processes. Characterization of blending in agitated vessels is usually by means of blend time. This is the time re- quired to achieve some specified degree of uniformity afier introduction of a tracer. The various techniques are summarized in Table Il. Each technique measures, a different degree of uniformity so that the time re- quired for blending may vary from one method to the next. However, the principles of correlating blend time, as derived from dimensional analysis of the relevant twansport equations, arc applicable to all techniques. Correlation method for blend time ‘The result of the analysis for equations of motion and mass transfer can be applied directly to the corre- lation of blend time. For negiigible gravitational effects and fixed initial and boundary conditions, we find from Eq, (12) that the concentration distribution within a vessel is a function of time, Reynolds number and Schmidt number. Experimental measurements for blending are not made of the concentration distribution itself but of the time required to reach a particular concentration distribution. Dimensionless blend time, tp", becomes the dependent variable in such a case, and Eg, (12) may be restated as (NresNsc) (28) Although molecular diffusion is the final step neces- sary to complete the blending of miscible fluids, it is not significant relative to convective effects in det mining blend time. This eliminates the Schmidt num ber as a correlating variable. Therefore: fy" =S pe) @) In a manner completely analogous to our previous discussions for power and velocity, we can show that dimensionless blend time is a constant at high and low values of the Reynolds number. The general form of this relationship is shown in Fig. 1d. Blend-time correlation Itis commen to combine some geometric effects the dimensionless blend time. Typically, for a given type of impeller, the ratio of impeller diameter to tank. diameter is raised to some constant exponent, The em- pitical correlation has the form: WD/T! = S(Naw) 0) ‘The Froude number has been included in some car. relations of blend-time data. However, where no density differences ate present and the liquid surface is fat, this is not justified. Gray [2] has summarized much of the literature on blending and blend time, For pitched-blade turbines, the correlation of blend time (as measured by the acid-base indicator technique) is as shown in Fig. 4 [/2]. At Reynolds numbers above approximately 100,000, dimensionless blend time be- Plched blade atines Blend time measures the concen distribution in an agitated system comes constant. The degree of uniformity attained at the blend time can be considered as within 1% of the final, totally uniform concentration of a tracer that is added at time zero. ‘The next article in this series will appear in the issue of Feb. 2, 1976, and will cover the fundamentals dealing with heat and mass transfer, non-Newtonian fluids, solids suspension, gas dispersion, and scaleup. References . Gag L, Henley, TL and Fei J. Gy Ohm. Bag, Dee. 8, 1973, pine Un, ¥. Wand Gray JB 09 ‘Academie Pra, Noor 8, 8. Nagata S, "Mining, Penile and Application! Haled Pres, key, New Vent 1975." Pee 8 Arwl 4 Churchi’S. “The Inerrttion and Ue of Kate Daa: The Rote Goaepi,* Sepia Puishing, MeCinw Fil, New York, 197 5 ed, RB, Stewac, W- Ean Light, EN, "Transpor Phnome ‘whe Ne Yor 100 1 White, AM. and Beemer, Tie. ACRE, Vo. 90, S85 (94), 7 Bate RL, Ponty, PL and Carpcn, RR, nd Boy Cn Pa Feo ane Pace” Vo inn Dip, Wat'2 No. 4510 aah 1 Rushon] Hi Conch EW. and Evra Jy hen. By. Pg, Aa Tain hoe Sopp. aan 9. Holland, FA ad Chapman, S. Ligid Mixing and Procoing in Sra Tani Rethel, New Von, 196, 10, Py, R.H. and Chilo, CoH, “Chenical Eagncos Handbcok,* 4th ea, McGraw-Hill, New Vork, 1568; 3h ad, 1978 1, Nagata 8 and other, Alem Rac Fag yn Vo 21, 260 (859) 12 Reale, Gand Fondy, FL, peed Ansa Meeting of AICKE, inne Gy Be 190, ‘The authors David 8, Dickey senor develope tar fr Chemis, ine PO. om “Bojgon, OF do4il Hee ‘i Seen at is oe seein re os wt iy ee ene eer ey ‘Jn eto gy ae Grecinsiyn, 13 Fundamentals of agitation The principles of heat transfer and mass transfer are useful in evaluating turbine-agitator performance even though proper design depends on principles of fluid motion, David S, Dickey and Richard W. Hicks, Chemincer, In Cl This article reviews the several correlations of heat id mass transfer for liquid-liquid, gas-liquid and solid liquid systems that are to be agitated. Also discussed will be heat transfer through a jacket or coil; fluid particle dynamics and settling velocities for solids; gas dispersion, including bubble breakup and surface areas and the effects of non-Newtonian fluids on agitator performance Heat transfer Liquid motion produced by a turbine agitator can be used to increase the effective rate of heat transfer. Heat ‘may be added to or removed from the process fluid by contact with a heated or cooled surface. The surface configuration and the agitator operation both influence the rate of heat transfer Heatransfer surfaces commonly used in agitated vessels are shown in Fig. 1. Either the tank walls or immersed tubes provide the surface area to contact the process liquid. An appropriate heat-transfer fluid is supplied to the jacket or tube in order to control the process temperature. ‘The tank jacket (Fig. 1a) may cover just the sides, or both the sides and the bottom of the tank. This jacket ‘can be a complete second shell on the tank, or a coil integrated into the tank wall. The helical-coil configu- ration (Fig. 1b) involves one or more oils of tubing mounted coaxially to the agitator shaft. Tube baffles (Fig. 1c) serve a dual purpose as the heat-transfer sur- face and as tank baffles, Numerous experimental con: figurations similar to industrial equipment have been investigated and reported [2]. Heat flow into an agitated vessel is determined by the temperature difference between the process liquid and heat-transfer fluid, (2, — #)), the heat-transfer area, Aj, and the overall heat-transfer coefficient, Uy As, ~ 9) 0) ‘Temperature difference in Eq, (1) is controlled pri- marily by process conditions. The heat-ansfer area in contact with the process liquid depends on the geo- metric configuration of the heat-transfer surface. The overall heat-transfer coefficient, Uj, is influenced by geometric and operating parameters. If fouling is ne- glected [3], contributions of the operating variables can he separated into three heat-transfer resistances: a= Yep Ti? nN) + GCA) In Eq, (2), the insice-film coefficient, fy, is inside with respect to the tank, or on the processiiquid side of the surface. The determination of this coefficient will be emphasized in the following discussion because it is usually the limiting resistance and is affected by agita- tor operation. Also in Eq. (2), the second term involves conduction through the walls of the tank or tubes, and the third term is the outside-flm heat-transfer coeffi cient. Dimensional analysis of the coefficient ‘The inside heat-transfer coefficient, fy, lumps to- gether convective and conductive effects in the uid 4 SETA EERE TERROR is Heat tansterf | _Hesicat eo 1 Hast transfer modiue medi Jacket (second set) Tube battes Heat-transfer surfaces in agitated tanks may be the actual walls of the vessel or immersed tubes Fig. film immediately adjacent to the heat-transfey surface. ‘This film coefficient is defined in terms of the total heat flow, 4, divided by the heat-transfer area, 4,, and the ‘temperature difference between the inside wall and the bulk of the liquid, (46), 4 A (MB), @) A sound basis for understanding and correlating the heat-transfer coefficient can be established by dimen- sional analysis of a physical model. The heat flow in Eq, (3) can be expressed in terms of the temperature gradi- ent at the inside wall of the vessel [4] = ffevelas eSferoa, ‘Combining Eq, (3) and (4) in dimensionless form by using Ad, and impeller diameter, D, as the characteristic temperature and length, respectively, yields “Pe five The dimensionless group on the left side of Eq. (5) is 2 Nusselt number for heat transfer in an agitated tank, From dimensional analysis of the energy-balance equation (discussed in the previous article [/]), the dimensionless temperature distribution, 0°, was found to be a function of Reynolds and Prandtl numbers: 6 (ah yh at A) = SNe Nor) 6) For given initial and boundary conditions, we can express the Nusselt number as a simple funetion of the Reynolds and Prandtl numbers by combining Eq. (5) and (6): a ©) Nya =S(N pepe) a Correlations for heat transfer ‘The Nusselt numbers in agitated vessels involve the product of Reynolds and Prand:! numbers each taken to different exponents (2,3,5]. Hence, the simple fune- tional relationship of Eq, (7) becomes: Naw = (Np Np)” ® For turbulent agitation in baffied tanks, the exponent B = 2/3 and the expongnt y = 1/3. This is consistent With similar correlations for flow in pipes. ‘When the temperature difference between the bulk liquid and the tank or tube wall causes a significant variation in viscosity, the ratio of bulk viscosity to viscosity at the wall, taken to an exponent, can be included in the correlation. The exponent for the viseos- ity factor has been found by various investigators to have a value between 0 and 0.4 Heat-transfer correlations for agitated tanks rarely use the impeller diameter as the length dimension in the Nusselt number [2,5], For geometrically similar systems, selection of the characteristic length is arbitrary and a matter of convenience. However, intuition says that the length dimension appearing in the Nusselt. number should be representative of the heat-transfer film thick- ness. For a jacketed tank, this thickness is some fraction of the tank diameter. ‘The same basic correlation for turbine agitation of a Jacketed tank wasestablished in three separate investiga~ tions [6,78] CF) Impeller diameter Duy Diffusion coefficient | |g" Panidle diameter |G ube diameter | A Drag force | e° — Acotleration of gravity (magnitude) | fe Gravitational conversion factor | AE Heattranster cotieient for helical coil (process side) 4 Tesi (proces side) hea-transfercooficient 4, Heateransfercoeficient for tank jacket (proc: es side) ‘hy Outside (transfer-fiui side) area of heat-trans fer surface sh, Hleatetransfer coefficient for tube baffes (proc: cs side) K Constant, Eq. (20) Thermal conductivity of liquid 4h, Liquid flm mass-transfer coeficient i Thermal conductivity of wall Length M > Masttransfer rate per unit volume mm Constant of proportionality for power-law fluid, Eq. (28) N Agitator rotational speed Nye Nusselt number, Eq. (5) Np Prandtl number Nps Reynolds number peg) Particle Reynolds number Max” Sherwood number, Eq, (19) Weber number, Ba. (22) | Nwe 1" Exponent for powerlawefsid, Eq (28) ty Number of tuber bales B Actual power requirement in gasied agitator 2, Agitator power requirement (without gas prox | en) 4. Rate of heat transfer &, Valumeaie Row of gas So Surface area i 5+ Dimensionless surface area | 1 Tanke diameter | {5 Dimensionless time U, Overall heattransfercoeficient ty) Realive velocity between particle and uid Superficial gas velocity, Ba. (23) Terminal setting velocity ' Weocity | 4 Concentration in the lid at gas interface 2 Concentration in liguid phase XE Gencentration in the liquid st solid interface Dimensionless position components Liquid level Arbitrary coefficients and exponents Heat-transfer-surface wall thickness Temperature difference between wall and process liquid Temperature of process liquid ‘Temperature of heat-transfer fuid Dimensionless temperature Vis00 Viscosity at wall conditions Density Particle density 1 Surface tension Shear stress Differential operator Dimensionless differential operator e pore ae ‘A more general relation can be written for the jacket ‘Nusselt number (7,¢]: (oo) ")" aren Correlations have also been developed for helical-coil configurations (2). Adding a coil of tubing for heat transfer drastically increases the problems of maintain- ing geomettic similarity. In addition to the length di- mensions associated with the tank and turbine, there must be included the diameter of the coi, the diameter of the tube, the length of the tube, and the spacing between coils, One correlation for coil heat transfer was developed by Oldshue and Gretton [9]: (28) =a 22" arg o the Nusselt number for the coil heat-transfer coeffi tion developed by Dunlap and Rushton [/0] for rows of (2) = on 238)"(2)" aree” Eq, (12) includes the effects of geometric changes for the impeller-to-tank diameter, and changes in the number of baffles, but neglects effects due to changes in tube-to- tank diameter. 16 —SHRRTOAT EROREERINE FEBRUARY = Design aspects for heat transfer Heat-transfer correlations are only as accurate as the physical and operational properties can be determined, Accurate information about fluid properties is not al- ‘ways available, or properties may change during opera- tion of the process. Thus, some degree of conservatism is required in estimating heat-transfer coefficients. In ad~ dition, the procoss-side heat-transfer coefficients deter- mine only a portion of the overall heat-transfer rate. Heat transfer by conduction through the tank or tube walls and the film coefficient in the jacket or coils must also be determined. Typical tank-wall or tube-wall thicknesses and film coefficients for condensing steam have relatively little effect on the overall coefficient. Overall heat-transfer to cooling water is somewhat more sensitive to the conditions in the jacket or coils. Addi- tional resistance to heat transfer caused by deposits and fouling must also be considered. Although the heat-transfer coefficient depends on the degree of agitation, sizing an agitator to achieve a specificheat-transfer coefficient isimpractical. The coeff- cient is relatively independent of agitator speed. For example, doubling this speed will increase the heat- transfer coefficient by a factor of 1.58 (ie., 2) but Power requirements will increase by a factor of 8 (ie, 2) A reasonable approach to design isto select an agita- tor that provides adequate bulk-liquid motion, and then to alter the heat-transfer area, temperature driv- ing-force, or other process conditions to give the desired e= Suspension of solids The problem of solids suspension is considerably mare complex than that involving single-phase liquid motion, since a second phase is moving in the tank. ‘When the solids are small and approximately the same density as the liquid, the particles move as part of the liquid, and the mixture behaves essentially as a single- hase liquid. However, when the solids settle more Partial suspension bs. Complete suspension «Unite suspension (some slid rext on Bottom (all solis are off the bottom) (solids supandsd ‘of the tank for short period) throughout tho tank) Suspension of solids in a liquid phase will depend on process requirements and properties of the solids Fig. 2 rapidly, sufficient agitation must be supplied to keep them suspended. ‘Agitation requirements in solids suspension vary, depending upon the process requirements. In some cases, the solids need only to be swept from the bottom Of the vessel to prevent them from accumulating. In other cases, a relatively uniform slurry is desired. When solids dissolve into the liquid phase, or when 2 solid- liquid reaction occurs, mass transfer must take place between the solid particles and the bulk liqui Development of design logie for selecting an agitator to provide a given degree of solids suspension requires some knowledge and understanding of fuid-particle dynamics. The additional degree of complexity, intro- duced by the presence of a second phase, limits the usefulness of the dimensional-analysis approach for the physical model, although this has been used for liquid motion. However, we can obtain some insight into this problem by examining solids behavior in a flow field First, we will look at the problem of a solid settling in a stagnant fluid, and then we will consider the com- plicating effects of agitation. Settling velocity Any solid particle moving relative to a surrounding fluid is acted upon by a drag force. The magnitude of this force can be computed from: Fy = OAs Pte as) ‘The dimensionless drag coefficient, Cy, in Eq. (13) is somewhat analogous to the agitatér power number. ‘The value of Cy is a function of the particle Reynolds number—Nagi) = dy tp /tt. This value has been*deter- ‘mined empirically for spherical particles and other sim ple geometries [3] For values of Npyg) less than 0.3, the flow around the particles essentially laminar and corresponds to the Stokes’-law region, where the drag coefficient: Ca = 24/Neagy (04) SCAT ENGINEERS PERROTT Ww This result for the drag coefficient is similar to the analysis of the power number under viscous conditions. ‘At high Reynolds numbers (1,000 < Ngrgy < 200,000), the particle wake is completely turbulent. For these conditions, the drag coefficient is @ constant, and is known as Newton’ law: Cy = 044 (as) Newton’s-law conditions apply when fluid inertial forces or form-drag dominate; these are similar to tur- bulent power numbers. The drag coefficient in the tran- sition range is a variable function of the Reynolds number[ 3) By using the drag coefficient, the terminal settling velocity can be calculated from Eq. (13), where the magnitude of the drag force is equal to the gravita- tional force resulting from a density difference between the solid and the liquid. In the Stokes"-law region, the value for drag coefi- cient in Eq. (14) leads to the following relation for terminal velocity 4 & a ey — 0/8 (16) Generally in the Stokes”-law range, the terminal veloci- ties for most, solids-suspension applications are slow ‘enough so that the fluid can be treated as a single phase for design purposes In the Newton’s-law region, use of the drag coefficient defined by Eq, (15) leads to a different relation for terminal velocity: 4 edt, — Oe an In the transition region (between the Stokes’. and Newton'slaw regions), calculation of settling velocity is an iterative process. Terminal velocities in the transi- tion and turbulent range will be handled by the solids- suspension design procedure. ‘The drag coefficient discussed here applies to spheri- ‘al particles. Correlations exist for drag coefficients of nonspherical geometries. These are of limited usefulness in design work, since detailed information on particle smetry is rarely available. Settling in an agitated vessel is more complex than setiling in a stagnant fluid, due to the presence of the turbulent-flow field. Attempts have been made to ana~ lyze the relative motion between a particle and an agitated liquid on the basis of turbulence theory, and some limited measurements have been made to deter- mine the slip velocity between the particle and the agitated liquid (1,12). ‘Although experimental results are limited, it appears that settling velocity is reduced, due to enhanced drag in the turbulent-flow field. However, no adequate de- sign correlation existing in the literature correlates the degree of suspension in terms of settling velocities under turbulent conditions Empirical correlations of agitation requirements for suspending solids exist [13]. Generally, these studies deal with measurements made in laboratory and pilot- scale equipment. By varying the solids size, density and concentration, and fluid viscosity and density, the ‘agitation required to achieve complete suspension ofthe solids is-determined. 18 ERICA RGIS PERAK Eo ‘Superficial velocity is a measure of surface area for mass transfer for dispersing gas in a liquid Design aspects of solids suspension ‘Typically, selecting an agitator for suspension of solids requires information about the physical proper- ties of the solids and the liquid. The degree of suspen- sion can range from (a) solids being periodically swept from the bottom of the vessel, (b) full off-bottom sas- pension, and (c) virtually uniform solids concentration throughout the vessel. These different levels of suapen- sion are depieted in Fig. 2 Mass transfer in solids suspension When the solids dissolve or react, mass transfer be- tween the solids and liquid must be considered. The rate of mass transfer, M, can be expressed as: M = ky ay (%, — 1) (18) ‘The convective mass-transfer coefficient, k, is analo- gous to the heat-transfer coefficient, hy, in Eq, (3). The driving force is a concentration difference, (x, — %;). Both transfer rate and surface area are expressed per unit volume. ‘When all the solids have been suspended, the effective surface area, ay becomes independent of agitation. Mass transfer can then be handled in the same way ‘as heat transfer. Therefore, the Sherwood number, Nyy for mass transfer can be derived in the same way as the ‘Nusselt number for heat transfer: Nox = by, D/Das aa) Dimensional analysis of the equation of motion and the component mass balance identifies Reynolds and Schmidt nusabers as important parameters (/]. There- fore, for geometric similarity, a correlation for the ‘mass-transfer coefficient should take the form: Ney =J Naw Neo) 20) Geometric similarity normally cannot be achieved in solids suspension. The particle size would have to be scaled up by the same factor as other linear dimensions ‘The approach of using particle diameter in the Sher- aeeeeet +. Flooded impeller Insufficient agitation during gas dispersion by # turbine impeller ereates a condition of flooding by, Moderate dispersion noatlooded) ‘wood number and impeller diameter in the Reynolds number is.often employed. But the arguments for this choice have limited validity Like the heat-transfer coefficient, the mass-transfer coefficient has been found to be relatively independent of the degree of agitation once total suspension is estab- lished. Changes in the liquid-film transfer coefficients are not a strong function of agitator operation. Gon- sequently, once the solids are suspended from the bot tom of the tank, the total surface area is exposed, To make a significant increase in the mass-transfer rate under this condition requires large power inputs for additional agitation, Gas dispersion Gas dispersion is normally done by injecting gas near the bottom of a tank below one or more turbine im- pellers. The object is to dissolve all or part of the gas in the liquid while maintaining a well-mixed liquid phase. In a sparged tank, the gas volume is small rela- tive to the liquid volume, when compared to tower-type operations. “Most problems involving gas dispersion are extremely complex. In addition to the gas being dissolved, cher’ cal reactions often take place in the liquid phase. Some reactions are biochemical, such as in fermentation, In order to design for the complicated processes tak- ing place, the basic mechanism of gas dispersion must be understood. The importance of the dispersion mech- anism becomes evident when the rate equation for mass transfer, M, is considered: M = hay (3, —%) 21) Although each of the separate factors in Eq. (21) depend on process conditions (and location in the agitated tank), the product of the mass-transfer coef cient, fy, and the surface area, ap, is controlled directly by agitator operation. However, like heat transfer, and ‘mass transfer in solids suspension, the liquid-phase transfer coefficient is relatively insensitive to the degree of agitation over a wide range of agitator operation ‘The purpose of agitation in gas dispersion is primarily ‘one of increasing the suiface area per unit volume. Bubble breakup and interfacial area ‘The best photographic evidence of the complex gas- dispersion mechanism is found in studies by van’t Riet and Smith [15], In these stadies, the investigators found that vortices form behind the blades of a radial-flow turbine. As the high circulation’ velocity in the vortex dissipates into the radial fiow of fluid, small gas bubbles are formed. Dissipation of energy as the vortex disinte- grates is a critical factor in forming gas bubbles. Dimensional analysis of bubble breakup in gas-liquid systems yields a dimensionless group known as the Weber number, Ny, = p0#L/o, which is a ratio of inertial forees to sutface-tension forces. In agitation, it is common to use » = ND and L = D to give a Weber number for agitation as: Ny, = N*D'p /o (22) Correlations for bubble size have been made in terms of the Weber number. An alternative approach using turbulence theory shows that bubble size is a function ‘of power per unit volume. Calderbank in Ref. [2] dis. cusses the various approaches in detail. The surface area produced for mass transfer is also a function of the rate of gas introduced into the system. It is common to use superficial velocity to characterize the gas flow: : ald ‘The reason for this approach can be understood from intuitive arguments and reference to Fig. 3. Two agi- tated vestels of different sizes are designed to produce cequal-size gas bubbles. If these bubbles rise at the same velocity, then the gas holdup and surface area per unit volume will be identical for the same superficial ve- locity. The dispersion in Fig. 3 is idealized, but does illustrate this relation graphically 23) —GENGAT ENONEERTG FERRUARY 7, WS 19

You might also like