NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS TECHNICAL NOTE No, 1795 APPLICATION OF RADIAL-EQUILIBRIUM CONDITION TO AXIAL-FLOW COMPRESSOR AND TURBINE DESIGN By Chung-Hua Wu and Lincoln Wolfenstein Lewis Flight Propulsion Laboratory Cleveland, Ohio Washington Poy Janvary 1949 :‘TECH LIBRARY KAFB, Nit ION 0144933 NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS TECHNICAL NOTE NO. 1795 APPLICATION OF RADIAL-EQUILIERIUM CONDITION TO AXIAL-LOW COMPRESSOR AND TURBINE TESIGH By Chung-Hus Wu and Lincoln Wolfenstein SUMMARY Basie gensrel equations governing the three-dimensional compres- etble flow of gas through e compressor or e turbine are given in ‘terms of velocity components, totel enthelpy, and entropy. These equations are used to determine the radial motion of gas through en exial-flow compressor or a turbine and the corresponding effect on ‘the redial variations of the state of gas between successive blade vows in the case of steady, axially symmetrical flow. Tho aspect ratio of the blade rov is found to be en important factor in the cal~ culation when the effect of redial motion 4s included. The usual method, which neglects the effect of redial motion, 1s shown to be good only for the limiting case of zero blade-row aspect ratio, that is for the case where the axial length of the blade row is moh Jarger then the redial length of the blade rov. A sinusoidal redial~ flow path is found to give the effect of redial motion on the radial variation of gas. state between blade rows as small as likely without any diecontimity in the curvature of the streamline ani is sug- gested for use in design calculations. ‘The equations are applied to investigate the maximm compatible mumber of the redial variations of the gas properties between successive blade rove that a designer is free to specify. The vari- ous ways of taking up these degrees of freedom and the different types of design obtained are discussed. A general procedure is given to calculate the characteristios of a compressor or turbine of any given type of design, taking into account the effect of redial wotion of gas. Numerical calculations made for two types of com- pressor and one type of turbine show that even in the case of non- ‘tepered passage, there is appreciable redial motion and that the corresponding effects are of significant magnitude and should be ‘taken into account in design.2 ACA TN No. 1795 ‘INTRODUCTION The design of a compressor or @ turbine hereinafter called a turbonechine my be divided into two phases. ‘The first phase con- corns the type of design to be used or the determination of the nost desirable variations of velocity and thermodynamic properties of the gas in planes norm to the axis of the machine betwoon successive blade rows. The second phase concerns the design of Yiedes that vill give the desired variations of velocity and other properties of gts in these planes. In the first phase, the conti- tion of redial equilibrium, that is, the redial component of the equation of motion, mst be used. The flow of gas ina turbo- machine is curvilinser; it is curved not only by the whirling motion of gas, but also by the redial motion of the gas. The equa- tion of motion then specifies the radial pressure gradient required to provide the centripetal force to maintain the curved flov. ‘In figure 1(a), a curved stream surface over four stages of a multistage turbomachine is shown and in figures 1(b) and 1(c) are shown its intersections with planes normal to and containing the axis of the machine, respectively. The radial pressure gradient Guo to the whirling motion of gas is always positive, whereas that ue to the radial motion of gas may be either positive or negative depending on whether the curvature caused by the motion is either inward or outward from the axis of the machine at the point of consideration. If the gas lying on a normal plane between two blade rows, such as station 1 in figure 2(a), 18 considered, the equation of motion gives (neglecting small terms) 2 apt 5 Se yee. e r ie where p and p are the pressure ani the density of the gas particle, respectively, r and 2 are the radial distance from ‘and the distance along the axis of the gas particle, respectively, and V,, Vg, and Vz are the redial, tangential, end exiel componsnts of the velocity of the gas particle, respectively. From this equation it is seen that the effect af the two motions on the radial pressure gradient is quite stmilar. The effects of the two motions are both proportional to the producy of the square of the velocity and the curvature (1/r and d°r/az°, respec- tively) involved. Even when the radial motion involved is sell, if the axial velocity is high and the blade-row aspect ratio is large, the second term in the equation is of comparable magnitude to the first and should be included in the calculation. In the 2 ar azseot NACA TN No. 1795 3 past, however, the radial motion and its effect were usually neglected (references 1 to 3). (Such methods of calculation will hereinafter be referred to as the “simplified-radial-equilibrium eyproximation.") In reference 1, a gonorel form of the Euler equation 16 presented but is not used later on in the through-flow analysis because the simplified-radial-equilfbrium consideration may be expected to hola sufficiently far downstream of a single row of blades whore the radiel velocity af the gas becomes small, For ‘this reason, this approximation is adequate to predict most of the experimental results available on the radial varlation of the gas state far downstream of a single row of blades (references 1 end 3). No satisfactory data or theory exist for the general case of a blade row within a closely spaced series of blade rows, as in 2 multistage compressor or turbine where the effect of radial notion may be quite significant. In the present paper, the effect of the radial motion of gas is considered in applying the radial- equilibriun condition to turbomachine design. In the ANALYSIS, the general equations governing the flow of gas in axial-flow turbomachines are developed primarily for the case of steady axially symotrical flow, which corresponds to the ‘Limiting case of an infinite mmber of blades. The blades are replaced in the calculation by an appropriate force field. A method of solution involving the use of the basic equations in finite-difference form 1s discussed. Expressions are developed for the use of a large number of successive axial stations and for ‘tho use of three stations for a stage in which an appropriate radlal-flow path is assumed. ‘The equations are applied to investigate the maximm mmber of radial variations af the velocities and other properties of the gas ‘thet a designer is free to specify in any particular design. ‘The various ways of using these degrees of freedom and the different ‘types of design obtained are discussed. In the last part of this report, a general procedure of cal~ culation for any type of design is given. It is used to calculate ‘two types of compressor and one type of turbine with the purpose of investigating the effect of radial motion on design calculations. This investigation, conducted at the NACA Cleveland laboratory, yas completed in April 1948.Po Se PP ts ° en zen 8 NAT TW No. 1795 SYMBOLS ‘Tho following symbols are used in this report: ‘ron aspect ratio of blade rov, (—+~ velocity of sound. Lift coefficient specific heat of gas at constant pressure specific heat of gas at constant volume differentiation with respect to time following motion of gas force acting on gas particles by blades per unit mase of gas radial component of F tangential component of F axial component of F fom of redial-flow path mass flow per unit flow drea perpendicular to axis of ‘turbomachine form of radial-displacement distribution ee enthalpy per unit mass of gas, ¢ + ) constant axial length of blade row (fig. 2) Mach number of gas mass of gas seot1035 NACA TN No. 1795 5 v s oda a) = os =I polytropic exponent of actual expansion or compression ‘process of gas static pressure external heat transferred to gas particle along its path of motion per unit mass per unit tine gas constant redius measured from axis of turbomachine moan redius of radial-flow path (fig. 2(a)) x coaeeme: entropy per unit mass of gas simplificd-redial-equilibriu approximation absolute stream temperature of gas time magnitude of T veotor velocity of rotor blades at redius r internal energy per unit mass of gas with 0° absolute as ‘base temperature magnitude of V absolute vector velocity of gas radial component of ¥ tangential component of ¥ axial component of V magnitude of W vector velocity of gas relative to rotor bladeNACA TW No, 1795 ‘tangential component of W 1035 from some fixed radial line : axis per unit mass of gas, to viscosity per unit volume of a y naxtmm radial @isplacenent over blade height 2 distance along axis of turbasachine « angle betwoon relative velocity of gas entering rotor and axis of turbamachine B angle between absolute velocity of gas entering stator and axis of turbonechine 7 ratio of specific heats, (c)/cy) rN radial displacement across rotor, (rz - 7) tat, 8 Gimensionless turning, (> 5, Lt a small-stege or polytropic efficiency 6 angular distance measured » viscosity of gas v Kinematic viscosity of gas, (u/o) t angular momentum about 2 eve) ° mass density ° Dlade solidity g @issipation of energy due gas per unit time ° function © angular velocity of rotor Subscripte: 1 in front of rotor 2 behind rotor and in front of statorNACA TH No. 1795 7 3 dehind stator and in front of next rotor 4 behind next rotor ° satiefying continuity equation ° satisfying radial-equilibriun and totel-enthalpy equations a at hub 1 any station 3 station short distance downstream of station 1 x any station in front of rotor L Limiting value 1 eny station in front of stator 2 sed vith r to indicate meen maius n used with r to indicate radius where maximm redial @ieplacement occurs 5 simplified-redial-equilibrium approximation + at tip ANALYSIS Basic Equations The state of gas in threo-dimensional motion 1s completely specified by its absolute vector velocity V, or its three cam ponents Vy, Vg, and V, referred to cylindrical coordinates x, @, and 2, respectively, and two thermodynamic properties. These last two properties are usually chosen as the static pressure p and the static density p, but herein it is more convenient to use ‘the total enthelpy per unit mass H and entropy per unit mass 6, which are defined by a Behe qa)8 NACA TW No. 1795 ds = du4p apt (2) For the range of temperature and pressure encountered in compressors and turbines; p, p, and T are accurately related by the fol- lowing equation of state p= pRr (3) ‘The Navier-Stokes equation of motion for a real fluid ie given, in vector form, by - doen [e742 vo] (4) sg ‘Tho energy equation for a real fluid is given by (refer- fences 4 and 5) “1 Dey p De g+h pete BH =a+k (s) The continuity equation can be written as #4 0H =0 (6) o veV 4 log, p 20 (6a) When the preceding equations are combined, the following four general equations are obtained (for derivation, see appendix A): wmeFemostx(vxt)-He vfrsivwn] u a+$ed 1B +o[v +2 wo] (8) rf Rf (2) Sle 310ACA TH No. 1795 9 (10) ‘Equation (7) gives the gradient of total enthalpy in terms of blade force, viscous forces, velocity, and other properties of the eas. This vector equation gives three scalar equations in three dimon- sions. Equation (8) gives the rate of change of total enthalpy of gas along a streamline in terms of rate of heat addition, rate of work done by blade and viscous forces, and so forth. Equation (9) Gives the rate of change of entropy along a streamline in terms of rate of external heet addition and of dissipation of energy due to Viscosity. Equation (10) gives the continuity relation in tems of velocity, temperature, and entropy of gas. It should be noted that of the six scalar equations given by equations (7) to (10) only five are independent, because they are derived from equa- tions (4) to (6), among which equation (4) yields three scalar equations, giving a total of five independent equations. (The te is used in the derivation of both equations (8) and (9). Steady Arially Symetric Flow Axial symmetry, that 1s symmetry about the axis of rotation, can be assumed to exist sufficiently far downstream of any blade row and is true everywhere for the Limiting case of an infinite number of blades. This assumption is usually made to simplify the analysis. ‘The blades are then replaced by a volume distribu- tion of forces, the magnitude of which is obtained by maintaining constant the product of the number of blades and the resultant force at any point on the blade as the number of blades is increased. (The resultant force is the difference between the forces on the two sides of the blade.) That 18, the resultant force acting on the gas by a blade element at any radius is con- sidered to be evenly distributed over the stream chest between two Dlades at that radius. Reference 1 chows that for incomress- ible and frictionless flow the value thus obtained gives an aver- age value with respect to the coordinate 6. Because the number of blades is usually large, this assumption is considered reason~ able for steady operation of turbomachines, and particularly for this invectigation. Thus all partial derivatives of gas prop- erties with respect to time and angular position 6 will be taken as equal to zero, The state of gas is considered ase function of r and z only, that is, the problem is reduced to ‘a two-dimensional treatment. With this simplification and byca HACA TN Wo, 1785 transforming V2 % into [v(v-¥) - x (vx ¥)], there are ara the redial, tangential, and axial components of equa- tion (7): av, ee Ye ay, Ye 20%) . ye Ve Baa, eede, Be 5, Me or or eo] & PT) a BE) 2 te Brae at 3 ee Vp (2¥9) Bie 1 M(r¥—) 4 W2Vg) Ve Oe aaa aa ap oats ae ase (7) and a, _ 4 MO) 4 yy My, My, (10) +? [an Ore +352 See * Le : In the radial equilibriun equation (7a), the relative impor- tance of the various terms depends mainly on the type of design. The radial force exerted by the blade depends on the twist and the taper of the blade in the redial direction, Tho redial variation of entropy depends on the radial variations of external heat sub- traction and of dissipation of energy due to’ viscous effect. In ordinary compressors and turbines, the amount of heat transfer 16 small por unit mass of gas flow, and its effect on the radial var- Aation of entropy is negligible. But in turbines with blade cool- ing, the heat transfor may be quite lange and it may give a signif- ieant redial variation of entropy, which must be teken into consid~ eration. The inorease of entropy due to dissipation of enerey bySeot NACA IN No. 1795 n viscosity is small and may be assumed to be the same at all radii over the wain portion of gas outside the boundary layers at the rotor drum and the inner wall of the outer casing. ‘The third tem on the right-hand side of the equation is equal to zero for froe- vortex type of design, in which rVg is set constant at all radii, ana has a nonzero value in other types of design, Similarly, the fourth term may have a nonzero value if constant axial velocity at all radii is not specified in design, The fifth term represents the effect of radial motion on the yadial-equilibriun condition. In the gap between tvo blade rows it 4s proportional to the product of the square of axial velocity and the curvature of the flow path in the arial plano and is very similar to the effect of whirling motion represented by V9"/r- If tho curvature caused by radial motion is positive, the effect is to decrease the redial gradient of total enthalpy or pressure caused by the vhirling motion of the gas. If the curvature is nega tive, the effect is to intensify this radial gradient. For design with large blade height, short axial blade length, high whirl, and high axial velocities, it will be shown that the effect of radial motion 1s large and should be considered in the calculations. The last four terms in the bracket are usually of the cane order of magnitude as the preceding volocity derivatives and ‘because they are miltiplied by v, which is much smaller than other multipliers (V,.or Vg), the whole product is much emaller ‘than the other terms in the equation and may be neglected. The same argument applies to similar terms in equations (7b) and (7c). Hence equations (7a), (7b), and (7c) may be simplified to aH de Vedlt¥_) _ av, oy, Bette a a Re wo a _ Ve C29) Ve or, - Fe ye (70) and mE arte re ls oe de et TE TRE OR ca2 ACA TN No. 1795 fmong the terme on the right-hand side of equation (8),other than the first term (the importance of which depends on whether there is blade cooling), V-F is the predominating term in passing through the rotor. Whereas in passing through the stator, all the torms on the right-hand side are of the same onder of magnitude. However, by the use of the equation of motion, the energy equation for steady flow, and the assumption that the heat generated from ‘the frictional work remains in the stream, DH/Dt can be expressed for steady axially symmetrical flow ina more useful form in terms D(z¥9) of only @, @, and —52 (seo appendix a): pre D(z¥9) Fe +o (8a) In equation (9), the two terms are of comparable magnitudes in ordinary compressors or turbines. With blade cooling, the first term is predominate. For the axially symmetrical flow, it reduces . weg tle (2) «2() 2B) «(ey «es BY [Pf ape atl ow For axially symetrical flow, equation (10) reduces to the (FB 2000 8+ te 3 toe wR()-"R(-° om none the proseding six equations, equations (72} to, (100), only five aro independent, because both equations (82) and ts} represent the energy relation. For nonviscous flow, there is an 1085 oo1085 NACA TH No. 1795 Bb additional relation in that the blade force is normal to the sur- face of the blade and consequently is perpendicular to the relative velocity of the gas, that is, F-(¥-H=-0 (a) ‘Then equation (8a) can be obtained by using the motion and energy equations and equation (11). (See appendix A.) Consequently, either equation (8a) or (11) can be considered as representing this additional relation, giving, in all, six independent equations (equations (74), (7e), (7£), (9a), (10a), and either (8a) or (11)). For such flow, equation (9a) reduces to simply Ds -@ eT (s») For viscous flow, equation (11) is no longer true because the foroe exerted by the blade 1s now inclined from the direction per- pendicular to the relative velocity of the gas by an amount to be determined from the shearing stresses of the gas adjacent to the Dlade surface. Under the present consideration of axially symetrical flow it is found desirable to retain equation (8a) as ‘the enorgy equation and to obtain the entropy change from H, V, and the polytropic exponent n of the actual compression or expansion process: pe = constant (12) Then the rate of change of entropy along the streamline is (see appendix A) Be am aay (Ye B toes + ¥, & tose 2 (23) De fa-1) (7-1) \"e Sz 1°82 2 Sz In equation (15) n is considered as known. In a given machine, n may be obtained directly from measured preseure and temperature data, Ina new design, 2 may be obtained from the assumed poly- tropic efficiency used in design calculations: For the compressor or one 1- stele a 7a NACA TH No. 1795 ‘For the turbine or one 1035 1 Because the change in s is usually amall compared with the changes in H and V, the preceding method of determining s seams to be adequate to account for viscous effect in calculating the pressure and density change along the streamline for the present problen. Method of Solution If efensions (2), (2), ana (3) are considered as the relations and? in terms of H and s, then equa- Hons’ (1a), Ue)s’ (7), (Ba), (8b), ead (10a) are tx tnfopendont equations in the case of nonviscous flow, and equations (7a), (76), (72), (8a), (10a), and (13) are six independent equations in the case of viscous flow both involving eight variables Vz, Vo, Vzs E, 8, Fy, Fg, and Fz. “In the direct problem, in which the shape of the blade profile, the shape of the inner and outer wall cf the gas passage, the rotor speed, the power input, and the entering ani exit conditions of the gas are given, it is theoret- ically possible to determine the variations of these quantities ‘throughout the machine. In the inverse problem, in which the desirable variations of two of tho gas propertics are pre- soribed, 1t 1s also theoretically possible to determine the varia- tions of the other properties af tho gas and the blade force necessary to achieve the proscribed variation of gas conditions. However, it seems that no general analytical solution of these equations is possible in either problem. ‘Two mmerical methods of solution are therefore suggested. In the first method, the previous equations are expressed in finite-difference form and gpplied to successive axial stations that are a short distance apart. In tho second method, only three ‘stations are used for each stege in which an appropriate radial-flow path is assumed. Mothod of finite difference for successive axial stations. At each station, 1f Vg is denoted by £, equations (74), (7e), and (72) can be written as :WACA TN No. 1795 1s By, | BY, Barer. 4 Mov, (%-%) (ve) (7a) and, ov, av, one y,(Fe- Fe). 4B (74) Between any two successive stations 1 and J that area short distance apart (fig. 5), the change in total enthalpy is - given by equation (Sa) 3 Hye) - mood) w o[Byley) - Hay] + flee (on where (r) indicates that the gas properties at a particular sta- tion are & function of the radial position of the gee particle in ‘that station, (It shoula be noted that due to radial motion the radial position of a gas particle at any station ) 1s different from its medial position at the previous station i.) In passing ‘through the stator, w = 0 and there is no change of H along each streamline except from heat effect. The entropy change between the two stations 1s obtained from equation (15) (see appendix A): vie ay(ry) - ory) = R Bay log, (asa) a3 (ne 1) (y~2) v2 fi By- Instead of integreting equation (10a), the continuity relation ‘between the two stations is readily obtained by equating the mse flow at the two stations. Gy ry ary = Gy ry ary (a4)16 NACA TW No. 1795 By expressing G in H, V, and s, equation (14) becomes (see appendix A): %2,4 ( ; +) ‘Equations (7g), (7h), (71), (8b), (15a), and (14a) are now six independent equations relating the gas properties and blade forces at the two stations. In these oquations, the heat-transfer term ie negligible in ordinary turbomachines and can be estimated in the case of cooled blades; the temperature T 1s a known function of H and V; and rj ie obtained from r; and Vz,4. Hence there are only eight tnknowns in Hy, 83, Vr,3» Vo,g» V2,37 Fr, gr Fo,j) and F,,4 (those at the first station 1° are considered as known). For a given blade operating at a given speed, two additional relations are known anong the velocity components from ‘the tangent plane to the blade surface at the point (rj, 25) Decause the flow of gas has to conform to the shape of the blades. Yor a new design, tho desirable variation of two af the eight variables (usually one of the two is V,) can be specified and the remaining five determined from the pregeding equations. It is to de noted, however, that, in practice, tho radial blade force is not essontially an independent variable to be specified by the designer, but is mainly determined by the actual construction of the blade to mest the aerodynamic as well as the mechanical- strength requirements at different radii. The designer has to see to it, of course, thet the redial force to be obtained from the blade actually constructed is consistent with that used in or obtained from the design calculations. The procedure of calcula- tion for this step-by-step method varies with the type of design, ‘the conditions given, or the two gas properties prescribed. The calculation is quite laborious and seems to be justified only in the process of actual design. (14a) In order to obtain en over-all picture of the radial motion in a turbomachine and its effect on design considerations, the fol- lowing method of ueing only three stations for each stage with pre- scribed radial-flow path may be used.NACA TN No. 1795 a7 Mothod af prescribed redial-flow path. - In a turbomachins, the radial motion of the gas ie caused by three factors: (1) Papering of the annular passage either at the inner or outer wall gives the flow a radial displacement across the stage, which 1s, of course, greatest in the inmodiate neighborhood of the tapered. surface. (2) Bven vith @ nontapered passage, a redial dteplacement across the stage may be called for because of a variation in the distribution of specific masa flov over the blade height. (3) Bven 4f there 1s no radial displacement across the stage (that 4s, the same particle occupies the same radial position at the firet station of each successive stage), there will, in general, be radial displacement of flow within the stage. This radial flow will then be oscillatory in nature, a redial displace- ment in the rotor being followed by an equal and opposite radial @isplacenent in the stator. This radial flow arises because of ‘the difference between the redial variation of the specific mas flow within the stage and those at the entrance and exit stations of the stage. (This radial displacement can only be avoided by specifying zero or the same redial variation of specific mass flow at all stations of the stage in the design.) In general, the radial flow of gas therefore consists of @ gradual, generally monotone, radial motion due to factors (1) and (2) with an oscillatory motion af period equal to the stage length due to factor (3) superimposed on it, The radial flow caused by these three factors will be. similar to that shown in figure 1, The effect of the redial motion on the calculations arises chiefly ‘through the term ov,/dz in the redial-equilibrium equation (7g). This tern 1s expected to be significant mainly because of the oscillatory motion, which may require significant changes in V; within a single row of blades. The case of oscillatory motion within a stage with no over-all redial displacement across the stage Will therefore be considered first. That is, the gas passage is nontaperea and the radial distribution of gas properties at the entrance and exit stations of the stage is the same. Because there is no blade force acting on the gas and there is Little time available for the gas to mix, the gas flowing through ‘the gap between two blades is under a constant pressure gradient =~ end consequently tends to move with the same curvature it acquires while leaving the firet blade. For straight passages, the maximm end minim points of the redial-flow path are likely to beas ACA TH No, 1795 somevhere near the middle of the gap. (The intersecting curve of a stream surface with an axial plane is herein referred to as "the : radial-flow path." Because of axial symmetry, the radial-flow path 4s the sano in any axial plane.) ‘The stations between blade rows are most conveniently chosen at these points, ‘The stations in front of the rotor, between the rotor and the stator, and behind the stator are denoted by subscripts 1, 2, and 3, respectively. (See fig. 2(a).) If xr, and L represent the mean radial distance of the flow path and the axial length of the blade row, respec- tively, then the radial distance of the gas particle at position 2 is given by Fo m= - 2h a(2) (a5) at stations 1, 2, and 3 2=0,1,2 T % £(0) = £(2) = 1, 2) = -2 (as) £0) = £1(1) = £1(2) = 0 where f is a function giving the form of the radial-flow path and the prime indicates differentiation with respect to z/L, It fol- lows that ar 72-711 fz Wee Eo - 2p iy, (2) an at station 1, z= 0, We, = 0 ( imaamich as 3V,/8z is practically zero in passing through the gap. Similarly, at station 2, 2 = L, Ye,1 2 £"(0) (18)sgot TACA TH Wo, 1795 19 ]+;-y, 22 (200) ( Because £" (2/L) determines dv,/dz or the effect of radial notion on the radial-equilibriu condition, 1t is desirable that it vary continuously; this condition together with those of equa- tion (16) suggests £7(0) = ~ x2, 2"(1) = x? and equations (15), (18), and (18a) become, respectively, cos x 2 (19) 24] (20) (20a) For the sinusoidal form of f(z/L), the maximum absolute value of f£"(z/L) occurs-at z= 0, L, and 2b, and ts equal to x2, Even if f£"(z/L) is assumed constant between z = 0 and 2 = 1/2, thus minimizing the maximum absolute velue of f" in ‘the interval, the absolute value of f” quale 8. This assumption, however, necessitates a discontinuity in f" at z= 1/2. The values of x2 for the absolute values of £"(0) and £"(1) can be ‘therefore considered as small as is likely. The smooth variation of £%(z/L) and the minimization of the absolute value of f"(z/L)20 NACA IN No. 1795 at the stations are reasonable assumptions provided that the radial force exerted by the blade remains relatively mall. ‘In such case, ‘the sinusoidal curve is believed to represent the major harmonic of the actual redial-flow path, and the major effect of the radial notion may be obtained through the use of this simple curve, Tho radial-equilibrium equation (7g) may be written in terms of (rp = 1) by use of equation (20) wm a tl, ae = at fi +, Set. (at (ry - 2) 5 vy 3? ay a * 22 Hy Me ary 27 TL aes (21) where 11, 2,5 Q, When 14 4s replaced by the dimensionless variable ry" = & 7 =) ‘the dependence of this equation on the blade-row aspect ratio A is seen to be gh oeeot L ts a (4k vey 2h Gy RH, ay, + 2 a + (yt 3 (xt = my!) way, 4? (28) ‘Thie form of the radial-equilibrium equation is seen to contain a ‘term directly proportional to the radial displacement, to the square of the axial velocity, and to the square of the blade-row espect ratio. If the blade-row aspect ratio is large or the axial velocity is high, the effect of radial motion may be large even ‘though there 1s only a small amount of radial displacement across the blade row. This method is readily extended to the case where there is an over-all redial displacement across the stege due to tapering of the passage or due to variation in the design fron stago to stags. In figure 2(b), the radial position of a gas particle originally 1035seot NACA TH Wo. 1795 an et r) inotation1 1s at rs in station 3. For the oscillatory notion required within the stage, rz is not generally equal to 3 (my + 75). For the samo reason stated in the previous case, it i desirable to have the radial-flow peth consisting of a sinusoidal curve superimposed on the line passing through (zy, 14) and (zg, 5); that te oy, 1 ‘Lt 83) 2 -(F),- ot - BS inaemoh as dV,/d2 is practically zero in passing through the gap. With this value of dv,,/dz, the radial-equilibrium equation (7g) decomes any avy 4 al Tyt Fg) x2 2 a * yh dey CU" 3 (te ) 5 Vat (24) where 121,2,3 This equation is similar to equation (21). (If rz =7,, it reduces to equation (21).) A similar equation in dimensionless rj' can also be obtained for this case by dividing rj by (ry - ry)i- Atany point (z, r) within the blade region, the magnitude of F, consistent with this sinusoidal radial-flow path is obtained from 6quations (7g) and (25):22 NACA TN Wo. 1795 an de bat, Me en By, eto tee 1 eleta2)\ x az xz Vz +3( “Za ) Ev, (te fom Es om EE (2s) Degrees of Freodon in Design Before starting to design a turbomachine it 1s necessary to decide how the velocity and other properties of the gas should vary radially at successive axial stations to give the best type of design for the particular application, It is therefore necessary for the designer to know beforehand the maximm compatible nuuber of such variations that can be specified. In the discussion entitled "Method of finite difference for successive axial stations” it is seen that theoretically radial variation of two gas properties cen be specified by the designer at each axiel station within the blade region with the blade forces determined accordingly. In the free space between blade rows, the properties of gas remain constant along the streamline end no arbitrary change can be specified. In current design practice, the usual procedure is to specify the desirable gas conditions only at stations between blade rows (as well as before the first and efter the last blade row) and then either to select some standard blade sections or to design them on the basis of prescribed veloc- ity distribution to achiove the desired change of gas state across the blade row. In eithor way, one degree of freedom at the second such station is taken up in going from the firet station to the second station, The designer is therefore free to specify only one condition at each station throughout, the machine, with the exception of one station, usually the entrance station to the first stage. Ways of specifying degrees of freedam. - In this discussion the following two fundamental equations Will be used: In the stations between blade rows, if the radial variation of entropy is negligible, the radial-equilibrium equation (7g) reduces to au, ty ats av, 4 ay, Bye te ee et), (es) oe 1035NACA IN No. 1795 23 Equation (Sb) is applied to three successive stations of @ stage for the case where there 1s negligible heat transfer and is aiffer- entiated with respect to r, moos 2 at,\ dH, dr, dH, dr, glee ee (27) te, &° G,)° a, eae ‘Tho following ways of taking up the degrees of freedom at these stations between successive blade rows are discussed. (2) Constant work per unit mes of gas flow over the blade height. This condition is usually specified in the design of & ‘turbomachine, It relates { after the rotor to its value before the rotor in the following manner £2 (re) = Gy) + re St Uae (28) or at, dr, at. 2 ea et. 28a Seo oe where ry) is equal to (2 - £1) at the blade tip and is also ‘akan a tee = £1) ab other radii, Constant work over the blade height gives constant total- enthalpy change over the blade height. If the velocity at the exit of a stage is equal to that at the entrance, this condition also gives constant static-onthalpy change over the blade height. Under the condition of constant work, equation (27) redutes to @ GG "at (21a) Sr (29)2h NACA TN No, 1795 This condition holds for the first stage of a compressor and will hold for all succeeding stages if constant work per unit masa over the blade height is employed. If a nonzero value of dH, /ér, is desired, an initial preparatory stage must be specially designed to obtain this value. In the last stage, however, it is usually desirable that aH/or be nearly zero. (3) Free-vortex-type distribution of tangential velocity: z =0 (30) or Seen Yea (502) This condition is commonly used in turbines and compressors. For incompressible flow, in addition to this condition, constant total enthalpy and constant axial velocity over the bladé height can bo specified. But for compressible flow, radial motion exists and only one Gf the tvo additional conditions can be obtained in con- Junction with equation (30). (See NUMERICAL EXAMPLE AND DISCUSSION.) (4) Symmetrical velocity diegram, If ¥,,1=Tz,2 and r, "2, =1, the symotrical velocity diagram gives Yo,1 + Yo,2 =F (31) or ty +, =wr? (31a) Differentiating with respect to r yields: ab, aby ete te (31) te rpdrg or Vy. 7 V,,27 the symmetrical velocity diagrams may be defined by £, (@) +82 (2) = or, (32)HAGA TN Wo. 1795 28 Then af, af, te ey (2a) If Vo,3 = Yo,1» 84 Vz,3 = Vz,2 = Vz,1» the symmetrical velocity dia- gram gives very similar flows through the rotor and the stator, experiencing the seme turning. The change in static pressure or enthalpy is the same in passing through the rotor or the stator and the stage ig therefore often referred to as the "50-porcent, reaction stage." Reference 6 shows that the blade-profile lose 4g @ minimm with the symmetrical velocity diagram if the dreg- Lift retio 1s constant. (5) Constant axial velocity over the blade height: a, me (33) At very low speed of gas flow with no change in density, the spocific nase flow is also constant over the blade height; therefore, there ie no redial flow across the blade row and equation (26) reduces to ° (34) And in case ‘then, ' Tho equivalence af equations (38) ani (34) breaks dom, however, for current aircraft applications, in which cases the speed of gas flow ia high. If equation (34) is substituted into the radial-equilibrium equation (26), there 18 obtained26 ACA TN No, 1795 Be FE oo (34a) ‘The left-hand side of equation (34a) is the tangential component of fluid rotation Vx ¥; thus equation (34) 4s e condition for potential flow in the free space between blade rows. If it is desired to correct for the affect of the boundary layers at the inner and outer walls of the gas passage, instead of equation (33) an appropriate axial-volocity variation close to the actual one may be prescribed in design: oY, se =o) (52) (6) Constant specific mass flow over the blade height. In order to avoid redial movement across the blade row in compressible flow, it has been suggested (for exemple, in reference 7) that constant axial velocity be replaced by constant specific mass flow ao, a6, ee ears (55) Radial displacement can also be prevented by the use of two condi- tions instead of three 2-2 (ss) For designs using either of these two conditions, the simplified- radial-equilibrium calculation is more correct. ‘Designs employing no radial flow have the advantage that the calculation does not Anvolve any radial displacement across the blade row and that the ‘two~dimensional-cascade data can be directly applicd. Tho equa~ tions for tangential- and artal-volocity distributions derived fram these equations (equations (35) and (36)), however, are @iffioult to solve and the conditions ere incometible with tapered passage in a multistage turbanachine. (7) Constant Mach mmber relative to blade. For a fixed ratio of exit to entrance velocity relative to the blade, the temperature or pressure change of gas across the blade row at any redius is proportional to the equare of the entrance Mech mmber of gasseot NACA TN No, 1795 27 relative to the blade at that radius. It might therefore be desirable to reach the Mach-number Limitation at ell radii. That is, for the rotor, My? = Sy = ee (s7) a ra(a,- eater) 2 2 w2 Vz,1° + (ry, - Vo,1) ay? a where M4, is the limiting inlet Mach mmber for the blade. Dif- ferentiating with respect to r, gives (7-2) my? ae Qe 42) a (¥,, 249,12) +20 fos “ye (378) For stator (38) ay 2 2 +5 ( . teat * Yee") Differentiating yields (3) 2 = «(247% 42) Be aye? + Yoo?) (oa0) Equations (37a) and (36a) may be combined with equation (26) to eliminate H. (8) Constant turning. If the maximm work for a given size is desirable, the limiting turning value my be reached at all radii, which will give, of course, a redial gradient in totel enthalpy after the first blade row. This gradient may not be serious ina single- or two-stage unit, but it may not be desirable to use it for all stages of an oight- or nine-stage unit, For compressors, ‘the Limiting turning may be expressed by (reference 2)28 NACA IN No. 1795 Yo2-¥ a2 ~ Yo et = 100) (38) 'zy1 or Vo,2 au FO Vey co) where o is the solidity of the blade element at radius r. Dif- ferentiating yields Wp 2 arg Vo (9) In multistage machines, similar variation in either tan- gontial velocity, axial velocity, or specific mass flow may be specified at the similar stations of each stage (40) (41) or AG, _ AGs ars a GF Stages of multistage machines designed for similar variations of gas properties from stage to stage are termed "typical stages.’ ‘Types of design. - A large mmber of different types of design nay be obtained by different coubinations of those conditions specified in equations (26) to (42). These designs may be conven- dently divided into two groups. In the first group, the condition of constant vork at all radii is specified in the design. That is, equation (28) 1s specified, which gives: seo1035 NACA TW Wo. 1795 29 AS AS end (43) Sh | Be _ Hs Hh In cases where the symetrical velocity diagram is also specified, equations (52a) and (45) give and (44) Sy es & “Ss In the second group, the condition af constant work is not specified. The following tables present a few types af design in each of ‘the two groups. The way in which the degrees of freedam are taken up at each station and the known characteristics of each type are given. In the tables, subscript 1 refors to any station in the machine; k refers to any station in front of rotor, that is, stations 1, 3, 5, ...j and 1 refers to any station in front of stator, that is, stations 2, 4, 6,.... A typical stage may be considered as composed of either stations 1, 2, and 3 or stations 2, 5, and 4,ae sare on] tied ot oer] Gametertatioe of ony otase oath (BER) «0 ‘mt redial grodiaat in arial veloolty far comprowetoi Flow ESortant axial velootty over biate helaht for SSoompeooaible Fav an ees wo Ce 3 | ange negative radia} gredient of asia} velco yes en station eee oe Yea Pept HOE ftoter exttalgy soarvasos fron bed t0 Hy Behind rotor, axial velsoy decrees frm Bb 7 my 24 Taye Fy8 coo vou eee ‘Behind ctaton ial Yeloaity toaresoeo fox tab oy 1035 S6LT *ON ME VOVNceow 1 conta i Bee] Conditions | tno ware oon E ‘eal Specified at | ation spose tye ‘ooo ata | itled at cen cuareatertetice of any stage ’ Hes a, | Histat oe a i meee SS mation 5 I : 5 camrtent eoastant vate 1 ' axial can | ae ath _ ate g | veeotty | a f\ et. See ae In case af inocapress:bie flow, thts type & TFL] ae Sequin to tultel Tloy aorvea’blato zee at | ay aler iagitfiod-radday-oquiliorts cos” Oa ms) | eee Eidoresion thin tqye 22 evitrataat to ‘ 3 fae. oes ‘2 ‘firet typo in group { a ae : er Mae Tes { epee e dng { 1 2 ‘3 j 5 conmtant vee iso | 2 t min fiow | 2 a om i * ‘No mdial flow across rotor ami stator aay Bindan Sn noo of sontayened pesage r ay en? Initial stage ts necosmry :f estaring s |S. fa bao dstfecont medial gradters ta jae Sota enttalgy than ta regutzed } fama wack | ea, | seg | We rettal flow aaroée rotor ant otatar Bindos ia once cf uaatapered peomee } Initial guide vane te atocnasty storing exo te 0 dicfarart reddat gute i sj2.2 font of specific masa flov than Ja required 4 a Bs a \ | @ |crow 1 Bta-] Coatiticns | One wore oon ica) speottied et | aition epoo- "pe three ata- | ation at oor Guarctortstioe af any ctago tions of eny | tain station tage in machine 1s to meatal ao, 00 = SO, G2 Ads petacres| 7 | 7° |3r ovmtto lars ay = m0 Blade rove | @ presortbed Yo redial flow sorvas all. blades in on Sbroughout co value at caly| pins € ogui toe «of mataparel yaasege mching | 2 | ge =0 | cho station a 2 i tho pryortbot metas at machine anly ono station 12 s| geo ‘ho machtoe come | rmemina aeeee yel- 2 ea ee ey seer |. | aoe EE TEE 6 ue smc tein ojametaical ay, sideration for noutaperet paasuge, tan yeloosty | * ee gouvial velosities and vari dcae tro yro~ ding | a aty | ake portional to radius (whpel-type rotation) | (ae, * ay 7 am ‘and oguare af radius, respectively, (Soo ea laot part of appendix 6.) |= ee S6LT ON ME YOVALACA TH Wo. 1795 33 GANERAL METHOD OF CALCULATION ‘FOR GIVEN TYFE OF DESIGN In the previous section, examples are given to show how the different types of design are cbtained by specifying different con- ditions compatible with the degrees of freedom available. In the calculation of the variation of the gas state through a turbomachine of any given type of design, the effect of radial motion has to be included, in general, to get a more accurate value, The following procedure af calculation is suggested: Tho first step of the calculation 1s to obtain an approximate solution by assuming that the gas flows on cylindrical and conical surfaces for straight and tapered passages, respectively, and by neglecting the effect of radial motion on the radiel-equilibriua equation. This method 1s the usual simplified-redial-equilibrim calculation and gives an approximate solution for the case of very small blade-row aspect ratio. The gocond step is to obtain the solution for the two extreme Limiting cases of zero and infinite blede-row aspect ratio. If the Gifference between the two cases is not large, the eolution for the cage of a certain finite blade-row aspect ratio may be estimated from the two extremes. If the difference is found to be large, calculation has to be made for the finite aspect ratio. In the caloulation for the case of finite aspect ratio, the method developed using a sinusoidal curve gives exact resulte if such curve is prescribed in the design and is believed to give good approximate results in cases whore it is not presoribea in the design, Even with this simplification, a series of successive approximations is required. Simplif'ied-Radial-Eguilibrium Calculation In this approximtion, equation (26) becomes a teal, ang ay pete at (268) In the cage of nontapered passage, Vp =0, TL = rg =%3=7, and equation (26a) is equivalent to3 ‘NACA TN No. 1795 €) : Eee ae (20) When equation (26a) is combined with equation (27) by aby ay, ‘abe ab: bo af. ay, ae eek pee efeee| |e eee =e pe & te ey te (se Be) 5 By t Ye2 ay ts at, ay, "ates a Gm) Ts! 1 ad ‘1 For the case of constant work and nontapered passage, equation (27>) reduces to ay, ay, ag. Z,1 2,2 i ‘2 Val Ge = e,2 ee tz t1,t he Dt a3 = ats agy °%,,5 ar +e sar hae (27) ‘The radial variation of density and pressure 1s obtained from equations (26a), (A5), and (AS) (appendix A) for the case 92 = 0: 2 Bey? -2ap, 9.47 ay, ee oe ee (45) Pea Scie eee asad 2 Pet 9 pe de _ Vous! Oey 7 ee oy? a ae (46) P44” a ‘1 ‘i In these two equations, the last term is very small compared with ‘the next to last term and my be neglected, It becomes zero for the case of nontapered passage and the resulting equations can be nore directly derived, as is usually done, by taking the approxi- mation involved as the use ofseot NACA TN Wo. 1795 38 2 pisces by Oy Fy for the radial component of equation of motion in place of apy Vy 3? a) ay, i i 8,4 r rd By Gy ey 7 et ae), Vea ar For a given type of design and given design values of limiting Mach number and limiting turning, the variations of gas velocity at any inlet station of the stago are mown. By inserting the radial variation of tengential velocity into equations (45) ana (46), the radial variations of pressure p,(r,) and density ,(r,) exe determined. By coubining the variation of density with tho varia tion of axial velocity, the variation of specific mss flow G(r) is obteined. The total mass flow across station 1 is thon given by Lt tomy, ary FL, jh ‘The variation of gas properties at the next station (for example, station 2) must be such that the continuity equation is satisfied 7 2,% es f Brg Gy arg af Gary Gary Y2,n ‘Lh By assuming the value of a gas velocity, for example, Vz,2 at the hub, the velocity and density variations at station 2 canbe deter- mined in a similar mamer, and total mass flow oan be obtained. The correct distribution of V, 2, and so forth, which gives the correct value of total mass flow, can be obtained in two or three trials. When the results obtained are substituted into the contimity equation (BS) (appendix B), over a portion of the annulus a certain amount of rediel displacement across the blade rov quite different from that assumed in the calculation is obtained. The result determined by this simplified-radial-equilibrium consideration is36 NACA TH No. 1795 only close to tho case of zero aspect ratio of the blade row, which will be considered in the next section. This calculation can there- fore be used only as a first approximation to the zero-aspect-ratio caso ani also to give a starting value for the calculation of a finite aspect ratio. In appendix C, formas are given in dimen- sionless forms for thie calculation for two common types of design. Zero-Aspect-Ratio Calculations The continuity equation in its integral form (equation (BS)) determines only the function rz (rj) and not values of dV,/dz. wo limiting cases will now be discussed for which the evaluation of the term aV,/d2 is unecessary. If the blade row has an . axial length aufficiently great relative to the blade height (that 4s, if tho blade-row aspect retio is sufficiently small), the tem (Q¥,/dz); will be negligible in spite of any redial displacement required across the blade rov. ‘This extrene situation is designated the zero-aspect-ratio case and differs from the simplified-radial- equilibrium approximation in that the radial displacement acroes the blade row is properly determined and its effect on the state of gas is included in the calculation, For « constant-area gas passage without taper or with slight taper, this affect is small; conse- quently, a successive approximation procedure starting with the result of the simplified-redial-equilibriu calculation can be used. Thies procedure may be outlined as follows 1. Using the simplified-redial-equtlibrium approximation, find Va,1r Vz,2> tie ber Hy» Hes Gy and Cg as functions of ry. 2. Find rz (mj) by equation (85), using the value of G2 (1%) obtained from step 1, 3. Substitute these values of rp (xj) into equations (26) and (27) with the (3V,/dz) term equal to zero, to get a second solution for Vz,1, Vz,2» fa» f2r Hy, and He ae functions of ry. 4, Repeat steps 2 and 5 if necessary, using the value of Gp (r,) from step 3. In step 1 of each process, the same procedure as in the previous caloulation has to be followed. That is, the variation of gas seqt1035 HAGA TH Wo. 1795 37 conditions at station 2 is such thet it gives the same macs flow as ‘that in station 1. In the case where there is considerable taper at the passage walle, it is better to assume rz (r,) according to ‘the taper to start the calculation rether then to use steps 1 and 2. Infinite-Aspect-Ratio Calculation Tho other limiting case corresponds to a blade row with axial length negligible as compared with the blade height; this case is designated the infinite-aspect-ratio case. ‘The negligible axial length does not provide space for any appreciable radial displace- ment, hence for @ nontapered passage rz may be taken as equal to ry or G = Gp (47) and for a tapered passage, a relation similar to the one that fol- lows may be used: ra? ry? ut Fiat (47a) T2,t - Tin Ga = Because the change in axial length for a very small change in V,, is also very small, (3V,/dz); doos not vanish. Although ite absolute value does not affect the radial motion because of the negligible biade-row axial length, the relative value of doV,/dz in front of and behind a blade row is nooded to determine completely ‘the distribution of gas properties at these stations. If the load- ing of the blade is relatively symmetrical or is designed to give a sinusoidal radial-flow path, the curvatures of the radial-flow path at the two stations are equal in magnitude end opposite in sense. Then (), --@, co) In order to combine this relation with equation (26) in a simple mamner, it may be assumed that %2,2 &), =- V2 ), (49)38 WACA TW No. 1795 Then combining equation (49) with equation (26) at stations 1 and 2 ay, ay, a . Vega eh + Vy g Wee = Mee Ah ee (a0) 8 oe g For a typical stage of a given design, equation (47) or (472) together with either equation (48) or equation (49) will completely determine the variation of gas properties at the two stations. If the variation of gaa properties is completely given at ono station, equation (47) or (47a) alone completely determines the gas state at the other station. In eppendix D, formulas are given in dimensionless forms for two common types af design to calculate the variations of gas prop- erties Zor the two preceding limiting cases. The results so obtained will give the limits of the variation of the gas properties along the blade height. If the difference is large, it is vorth- while to make the folloving calculation for a finite blade-row aspect ratio. : Finite-Aspect-Ratio Calculation The results previously obtained for a sinusoidal radial-flow path is used. Inasmuch as there is no general analytical solution of equations (13a), (14a), (24), and (27) that is possible even for simple types of design, the method of successive approximation is used. The procedure of this calculation is somevhat similar to that used in the case of zero aspect ratio. In the case where a tapered passage or difference in design calls for en rz different from ZX), first ostimate the redial positions of gas particles at sta~ tons 2 and 3, rz and rg, respectively, as functions of 7. Then calculate the variation of gas ties at stations 1, 2, and 3 by equations (13a), (24), and (27), such that the total mss flow is the seme at the three stations. Using this result, find r2(y) and xg(r) dy the continuity equation (BS) and see if ‘they check the assumed values. By interpolating r2(r,) and xg(ry) obtained after a fow trials at different r), the value obtained by the next try will be close to the correct value. For the general case where Hj, Hz, £3, and 5 are deter- mined by design as functions of ry, a rough approximate solution gives (see appendix E)NACA TM No. 1795 39 (2-5) re-mset (s0) ‘This value can be used as a starting value for more exact calcula~ _ tion or may be used as the final value for approximate calculation, ‘This approximate value of radial displacement across the rotor is obtained in the following manner: First take as two separate functions be (x4) the function (rg - 14) of 1, satisfying radial- equilibrium and total-enthalpy-change equations (26) and (27), respectively, for a given distribution of the other variables Ae (m4) the function (rg - 7.) of 1, satisfying continuity equation (14) for a given distribution of other variables Tt ds assumed in this method that the radial gradients in V, and f depend primarily on the magnitude of the radial displacement ( 27%) and not on its emct distribution, Accordingly a (r) ia eet equal to de (ry) = yo 6 (7) (51) where y, is the meximm value cf 4, and g (rj) is 2 plausible form for the distribution of 4, satisfying the boundary conditions: & (r1,y) = 6 (r1,4) = 0 for (s2) @ (rn) = +e (ry) = If A¢ 4s caloulated for a couple of values of to, it is possible to plot yo, the maximm value of Ac, against jj. A fairly good approximate solution might be expected to correspond to the point Jo = Yo Tis process can be further refined by varying ¢(r,) from the function originally asstmed in the direction of the calculated function A,/¥q-40 NACA TH No. 1795 Distributions of gas properties calculated from rz (r,) given by the approximate relation (equation (50)) has been found to agree much better with the value obtained by tho previously described nothod of successive approximation than that calculated under simplified-equilibriun approximations. WOMERICAL EXAMPLE AND DISCUSSION The method of calculation outlined in the previous sections is applied to the typical stages of a compressor employing symotrical velocity diagram and constant total enthalpy, and a compressor and a turbine employing free-vortex and constant-totel-enthalpy design. The celculation 1s rendered dimensionless by expressing all veloc- ities in terms of U,, total enthalpy in terms of 0,2, and r in tems of ry, Bocause the main purpose af calculation is to deter mine thé magnitude af the oscillatory redial motion and its effoot on distribution of gas properties, a nontapered constant-aree passage is used. Heat transfer ie assumed to be zero in the calculation and the entropy is aseumed to be constant at cach station, The change of entropy across the blades at all radii is taken equal to thet obtained from the polytropic efficiency assumed at the mean radius. It then gives a radial variation of polytropic efficiency decreasing fron tip to bub. This variation seus to be in the same direction and of comparable magnitude with those obtained from experimental data. This caloulation does not take into account the boundary layers at the rotor drum and the outer cesing, and is consequently good only for the main portion of gas flowing between then. In the comparison of different blade-row aspect ratios in each design, in addition to the same aerodynamic limitations, the same axial velocity at the mean redius is used. The comparison Detween different cases will be slightly different if another basis of comparison 1s used. (1) Symmetrical velocity diagram and constant total enthalpy. - Because the difference between zero- and inPinite-aspect-ratio cases is found to be large in this design, calculation is made for a Dlade-row aspect ratio of 2. The design constants used for all cases are:NACA TN No, 1795 al Hub-tip ratdo oe ee ee OB Limiting Mach number relative to rotor blade... 2.2... 0.8 Ye,2 ~ Yon Limiting value of 22 Or Vz,2 Polytropic efficiency at mean radius «6. eee ee 9 Ve jrymlg cee eee eee ee ee ee OT (the last velue results from the use of Vz,1,n/0, = 0.8 in the simplified-radial-equilibrim calculation, and is used for all cases.) The results of the calculation are shown in figure 4. Figure 4(a) shows the distr{bution of specific mass flow in front of end behind the rotor for the different cases considerod. Tt may be seen that in all cases except the infinite-aspect-ratio case, the specific mass flow G/G, increases toward the hub faster ‘behind the rotor than in front of the rotor; in other words, passing through the rotor the gas moves toward the axis of the machine, The magnitude of this displacement 1s obtained from the continuity relation and is shown in figure 4(b). In tho simplifica~ radial-equilibriun calculation, it is assumed that there 1s no radial motion, but when the distribution of specific mass flow ia substituted in the continuity equation (B5), quite large radial displacement across the blade is obtained. ‘This kind of calcula- tion is therefore not a good one. In other calculations, the dis- tributions of gas properties are calculated from assumed radial Gisplacenents that are to be chocked with the displacements required from the continuity relation vith these distributions, and are therefore consistent in themselves. ‘The radial displacement to be used in the approximate caloulation for A= 2 4s obtained by the approximate formula (equation (50)) and is about 25 percent lower than the correct value. The variation af axial velocities is presented in figure 4(c), vhich shows that the arlal velocities increase tovard the hub in all cases but at different rates, The high value of exial velocity at the hub before the root allows the use of higher turnings throughout the blade height without exceeding the limiting value of (Ty o-Voi)Nz1 oF Oy, at the hub, It also helps to give a tore uniform fiach nuber relative to the rotor blade over the blade height. As a result, this type of design gives a higher pressure rise and a higher specific mass flow than a free-vortex ‘type of design using the eame design limitations, In order to utilize this advantege fully, the variation of axial velocity should be correctly determined.42 NACA TN No. 1795 ‘The calculation of axial velocity based on simplified radial equilibrium gives a result close to the zero-aspect-ratio case, - which is also true in the distribution of other properties in this calculation, because in the case of zero aspect ratio, the curva- ture caused by radial motion is negligible and the difference in gas properties caused by the redial displacement is very auall in ‘the case of a nontapered passage. ° ‘The variation of tangential velocities is shown in figure 4(4). These velocities in different cases vary ina similar uamner and the difference of magnitude between them is mainly due to the aif- ferent value of 64 determined by the different values of Yz,an/t, 3m the Tartous cases. Figure 4(e) shows the variation of air angles entering the rotor and stator blades. The difference between the simplified- radial-equilibrim calculation and the case of aspect ratio of 2 4s significant throughout the whole blade height. ‘The simplified radial-equilibrim calculation gives an error of about -5° at the Yip of the rotor blade and at the hub of the stator blades. ‘This difference results in an error in the angle of attack at the design point by that amount and the range of operation is also reduced. The variation of Mach mmber relative to tho rotor blades is shown in figure 4(f). Tho aimplified~radial-oquilibrium calcula~ tion gives nearly constant value whereas the more correct calou- lation shows that Mach umber actually decreases more than 10 per- cent toward the tip for the case of blade-row aspect ratio equal to 2. (This variation is only about half of that of a similar free-vortex compressor. ) Figure 4(g) shows the pressure distributions in front of and behind the rotor and the pressure rise across the rotor at dif- ferent radii. The difference in pressure distributions may explein to a certain extent the difference found between measurement and simplified-radial-equilibriun calculation, The pressure rise across ‘the rotor is fairly uniform in the case of an aspect ratio of 2 and is a desirable feature. ‘The velocity diagrams at three radii for aspect ratios of 0, 2, and « aro shown in figure 4(h). If this atago is dsed as the first stage of a compressor, the permissible tip rotor speed of ‘this dosign at standard sea-level conditions is equal to 868 and 826 foot per second for A=O and A= 2, respectively. Theoss NACA TW Wo. 1795 43 specific mss flow per unit annulus area corrected to standard sea- level conditions is equal to 41.5 and 40.0 pounds per square foot per eeconl for A= 0 and A= 2, respectively. (2) Eree-vortex and constant-total-enthalpy compressor. - The design constants used are the eame as in the previous calculation. In addition, Ve,1 and Ve,2 are considered to be equal to Wo,2 and Wg 1, respectively, at the mean redius. In this type of design, the simplificd-redial-equilibrium approximation is equivalent to the zero-aspect-retio case, because axial velocity 1s also con stant over the blade height due to the constant H and constant £ with respect to the radius and the negligible curvature effect caused by radial motion, That is, the same values of Hy, He, fa» tar Vz,1, and Vz,2 cour in both cases and the entire cal- culation is the same. (See aleo equation (E6).) Because the radial motion involved in this type of design is due only to the compressibility of gas, the difference between the zero- and infinite-aspect-ratio cases is not large; hence the cal- culation for a finite-aspect-ratio case is not made. The distribution of specific mess flow in front of arid behind the rotor is presented in figure 5(a). Even in the zero-aspect~ ratio or simplified-radial-equilibrium case with a constant axial- velocity distribution, there 1s considerable change in density, which requires an appreciable amount of outward redial motion to obtain the given design conditions behind the rotor. Although the amount of this radial motion is small (fig. 5(b)), its effect on the variation of gas properties is not entirely negligible. Its effect can be seen in the curves of figures 5(c) to 5(g), which are somevhat similar to the symetrical-volocity-diagram and constant-total-enthalpy design in nature but of smaller magnitudes. If this stage is used as the first stage of a compressor, the permissible tip rotor speed at standard sea-level conditions is equal to 758 feet per second for A= 0. This tip speed is about 13 percent lower than thet of the corresponding case of the pre- vious design. The specific mass flow corrected to standard sca~ level conditions is equal to 38.6 pounds per square foot af annular area per second for A= 0, which is 7 percent lower than that of the corresponding case of the previous design.a4 NACA TW Ho. 1795 (3) Free-vortex and constant-total-enthalpy turbine. - The design constants used in the caloulation are: M,<1, Uy/ay,4 = 0.5, Vor ,n/91,4 = 8) Vz,1,m/21,4 = 0-4, Vo,2 = 0, and polytropic efficiency at mean radius equal to 0.87. For the simplified-radiel- oquilibritm approximation or zero aspect ratio, V,;/0, 1s constant and so is V, 2/Uy, which is found by the continuity relation to be equal to 0.877. The same velocity at station 2 is used for the case of infinite aspect ratio, thus making the only difference at sta- tion 1, ‘he result of tho calculation are shown in figure 6. 1035 The distribution of specific mass flow in front of and behind the rotor is show in figure 6(a). Because of the constant axial exit velocity, the specific mass flow is constant behind the rotor. Except for thé case of infinite aspect ratio, there is an inward radial motion of gas in passing through the rotor (fig. 6(b)), the magnitude of which is about two and one-half times that in the previous free-vortex compressor (fig. 5(b)). Tho variation of axial velocity in front of the rotor 4s shown . im figure 6(c). An increasing axial velocity toward the hub of about 15 percent would be required for an aspect ratio of 2. Figure 6(a) shows the radial variation of gas angles entering rotor blades. The difference is only important at the hub, In the actual case of an aspect ratio of 2, the simplified caloulation would give an angle of attack 3° to 4° too high at the hub. ‘The absolute and relative Mach numbers of gas in front of the rotor are shown in figure 6(e). In the actual case of an aspect ratio of 2, the Mach number at the hub is about 3 percent higher than the simplified calculation. Figure 6(f) shows the pressure distribution in front of the rotor. For an aspect ratio of 2, the pressures at the tip and at ‘the hub are about 2 percent higher and 5 percent lower than the simplified calculation, respectively. The velocity diagrams at three radii for the zero and infinite aspect ratios are show in figure 6(g).ssot NACA TW Wo. 1795 45 ‘SUMMARY OF ANALYSIS AND CALCULATIONS In axial-flow turbomachines, redial motion of gas occurs decause of the tapering of the passage and the variation of gas conditions across blade rows specified in the design. The direc tion and the magnitude of this radial flow depend on the type of design, the tapering of the passage, the hub-tip ratio, the blade- row aspect ratio, and the speed of gas flow. Even in the case of free-vortex type of design employing a nontapered passage and requiring no change in velocity distributions from stage to stago, there is an appreciable amount of oscillatory radial motion within ‘the stage. This radial motion gives an additional tem to the ordinary redial-equilibrium equation. In the free space between blade rows, ‘this additional term is very nearly equal to the product of the square of axial velocity and the curvature caused by the radial flow. Depending on whether the curvature 1s positive or negative, the radial~pressure gradiont caused by the whirling motion of gas is decreased or increased, respectively, by this additional term. The determination of this redial-flow path requires a long process of step-by-step calculation, It is found, however, tliat @ sinusoidal radial-flow path gives an effect on the radial varia- tion of gas condition between blade rows as auall as possible without discontinuity in the curvature of the streamline. Inasmuch as it represents the major harmonic of the radial-flow path that may exist in any design in which the radial blade force is effec- tively small, the caloulation based on this simple radial-flow path gives good approximate results. It probably underestimates ‘the effect because of the neglect of higher harmonics, Tho analysis made of the maximum compatible number of the degrees af fresdom in specifying the radial variations of gas properties in stations between successive blade rowa of @ turbo machine shovs that under the conventional design procedure the designer is free to specify two conditions at any one station and ons condition at each of the remaining stations. The various ways to use up these degrees of freedom and the resultant types of design obtained are discussed. ‘The usual method of calculation, which neglects the redial motion, gives results close only to the case in which the axial length of the blade row is much larger than its redial length, and is not good for the case of a finite blade-row aspect ratio.46 NACA TN No. 1795 ‘The difference between the results obtained by the usual method and ‘the method suggested herein is found to be quite large ine design employing constant totel enthalpy and symmetrical velocity diagram. Calculation made for this type af compressor using the same limiting Mach number, seme limiting turning, same axial velocity at the mean radius, and for.a blads-row aspect ratio of 2 gives the following @ifferences between the usual and the suggested method: 1, The radial varlation of axial velocity in front of the rotor is 13 percent for the usual method and 28 percent for the suggested method, and the radial variation of axial velocity behind the rotor is 53 percent for tho former and 40 pércent for the latter (e11 expressed in terms of their values at the mean radius). 2, The air angles differ from 1° to 5° at the hub and at the tip. 3. The radial variation of Mach muber relative to the rotor blade in the former is 9 percent lover than that in the latter. 4, The radial variation in static-pressure rise across the rotor,is 13 percent for the former and only 2 percent for the latter. 5, The mass flow in the former is 4 percent higher than thet in the latter. 6. The permissible rotor speed in the former is 5 percent higher than that in the latter. - Lewis Flight Propulsion Laboratory, National Advisory Committee for Aeronautics, Cleveland, Ohio, October 14, 1948. seoACA TW Wo. 1795 a7 APPENDIX A DERIVATION OF EQUATIONS IN ANALYSIS From equation (3), the definition of h, and the relations au =o, at (a) Reo, =o, (a2) and _ ye (a3) 7 there te obtained a -ae() ws ‘From equations (1) and (a4) ame ( z an = Ba(B)ea ¢ (as) From equation (2) and the definition of h Tas = an - 32 (as) Uaing equation (A4) B)_ a tose dye(Q)-@ on From equations (2) and (3)48 NACA TN No. 1795 °@)- ae -Ae a (2086 2) (a2) a (160 3)- @ (loge p) (a9) From equations (4) and the following relations a8 -3.@07 (9) ¥ = 3 wv? - Tx (7 x) there is obtained Tx Wy x ¥) + dove - a Zwsek vs Eo (e-¥) (ato) From equations (AS) and (A7) Ly? 42 Rov? edo of =vH-T Vs Combination with equation (A10) yields ne ve eave 40x (7x0) - Hob [ve 42 v6 i) ol From equations (1) and (A6)NACA TW No, 1795 : 49 By combining with equation (4) iB, wr +e [a oS vom} (a2) When equation (All) is combined with equations (2) and (5) Beashidbev[ree v7 +E o(o-7)]} () ‘From equations (2) and (5) Ds angst or Beagand ) From equstions (6a), (3), and (A9), the following form of continuity relation is obtained: Prime 2-2 (geo 02) Equation (Se) is obtained by applying motion and energy equa- tions to e mass system with a fixed control surface as shown by the solid lines in figure 3. Under steady axially symmetrical flow, the mass infloy dm, in time @t 1s equal to the mass outflow amy in time dt, the state of gas within the control surface is not changed,and the state of gas at stations 1 and 3 1s constant with reepect to 6. By equation (A10), the sum of the tangential dlade force and the tangential viscous gas forces exerted by the surrounding gas particles on the systen is equal to50 NACA TN No. 1795 The torque about the z-axis exerted by these tangential forces on D(x Vo) the system is therefore simply —z-% am and the work input to the system by these tangential forces in time dt 4s equal to D(r¥p) FE amy ott = o[(2%e) 5 - (2¥9)5 ] am In passing from station i to station J, beside receiving this work input, the gas particles are doing work against the arial and radial viscous forces exerted by the surrounding gas particles. However, this negative work is usually small, and if it is assumed ‘that the heat generated from this frictional work is added back to ‘the gas stream then the heat addition cancels the negative work, ani the energy equation for steady flow gives a (ay = Hy) amy = f°? a am at + ofleve)s - (v9), Jom, ty or ne D(2¥p) PH 0+ oe (a) where Q denotes the rate per unit mass at which the gas stream sheet is receiving heat from external source through blades or other passage walls. Tt may de noted that for steady nonviscons flow, equation (8a) can be obtained by using equation (11). For euch flow, equation (8) becomes =Q4F-¥ sig Fron oquation (11) it can be seen that F-V is equal to ¥-U, However, F+U is equal to FgU or Fgrw and Fer is, dy equa- D(2¥9) ‘lon (Ta), equal to 5:2, Hence the preceding equation becones1035 ACA TH Wo. 1795 52 : Beato 20) (82) ‘When equation (12) is given, the entropy change can be obtained in the following manner. From equation (49) ana (3) Eat But by equations (12) and (3) ) = Fr Be (ose 2) - 3 (lowe 0) (a2) D a) De le P = 5 Be Se T Substitution into equation (Al2) gives GT H tose 2 (aa) For steady axially symmetrical flow, equation (A15) reduces to Ds ne 3 2 Ben gataty (eR ueet + 2 ime2) 08) For successive axial stations 1 and j a short distance apart, equation (A15) gives #5 (ey) - as Ge) = 8 Ey [Boe 25 (ry) - 2085 ty ()] Bz agg Ta (au) 1) (7-1) Se By ey) Inaemuch as the temperature change between the two successive sta- ‘tions is amall, the temperature ratio can be considered equal to the enthalpy ratio:52 NACA TN No. 1795 (13a) The density ratio between the two stations is obtained from equations (a9) and (3) ° J qlee (as) Py Combination with equation (A15) yielas a (a7) Substitution into equation (14) gives = 8 R ry ary = Ve & ssot1035 NACA TW Wo. 1795 53 APPENDIX B DETERMINATION OF RADIAL DISPLACEMENT FROM CONTINUITY EQUATION ‘Equation (14) may be written as a linear differential equation for 1,2 as function of r,, provided G,(r,) is known alg?) 2a) ay em) agg 2 a any ary + ney? (aay 1 vhon divided by rp 4? = \2 wo | am, ae ee) =~ G4) Baa © 1 If G, and Gz are not known, and only G/G,, end C/G are known, a modification is necessary oo ZY wa (Guay gt Se 72, ¢, T2,t/ S2,¢ r ‘1 yt (33) The value of G),4/G2,4 18 found by the condition that total mass flow at stations 1 and 2 is the same54 NACA TN Wo. 1795 2,h eye? Ge > (2) en T2,t) S,% oe =) he ay ev Hence TL,t ee) Tt At & j aie he (ay Be (y) ss) iy F2,%, Tat Gy (x) Lr & Trt T1,t ae (my) ThE Ay yy, et seotNACA TH No, 1795. 5s APPENDIX ¢ SIMPLIFTED-RADIAL-EOUILTERTUM EQUATIONS Equations to calculate distributions of gas properties under ‘the simplified-radial-equilibriun approximation for a few types of design are given. Free Vortex and Constant Total Enthalpy For this design =0, go =0 (c2) from equation (26a) in the section entitled "Simplified-Radial- Equilibrium Celculation," yt ceo (ca) The variation in tangential velocity is, by equation (C1), ae i,t Yo,1 = Ve,a,t Fy (c5) At each station, by using equations (46) and (C5), 2 Pt p72 Qo _ (Vo,t re) ma, ar 3 Py e When the preceding equation is integrated from r to ry, and the relation is used, there 1s obtained56 NACA TW No, 1795 ete WE 1 (10,t 7 bone (et) er co t This equation holds for all stations, provided the appropriate values of (Vp ;/a,) are used. Tt follows trom equation (C2) that at each station 2.2 Ge ee _ The radial position of gas at station 2 or 3 is then obtained by mmerically integrating equation (55) using distributions of specific mass flow given by equation (C5). An alternate method is to expand the right-hand side of equation (C4) into a binominal 2 Pr 2 series. Because 254 (es) (ie) . | is usually less than 0.15, three terms will be sufficient. Let 6, represent the average density in the annulus between r and ry, thenNACA TN No. 1795. 85 AEPENDIX ¢ SIMPLIFTED-RADTAL-BQUILIERIUM EQUATIONS a Equations to calculate distributions of gas properties under tho simplified-radial-equilibrium epproximation for a few types of design are given. Free Vortex and Constant Total Enthalpy For this design eto ate (cx) from equation (26a) in the section entitled "Simplified-Radlal- Equilibrium Calculation," ay, zi “Bi =o (ce) ‘The variation in tangential velocity is, by equation (C1), ry, . yt Vo, = %0,i,¢ Fy (es) At each station, by using equations (46) and (C3), 2 Pe 97-2 ap _ (Vo, re) aay, eo When the preceding equation is integrated from r to rz, and the relation is used, there is obtained56 NACA TN No, 1795 pry aie equation holds for all stations, provided the appropriate values of (Vp,,/a,) are used. It follows from equation (C2) ‘that at each station 2 Dada - 23 (2st Pe 2 \ay é-2 (cs) ‘The radial position of gas at station 2 or 3 is then obtained by mmerically integrating equation (ES) using distributions of specific mass flow given by equation (C5). An alternate method is to expand the right-hand side of equation (C4) into a binominal 2 2 2 sertes. seomne 7% ("2t) [(2)° — 2] se uruaay tose ten 0.15, three terme will be sufficient. fet p, represent the average density in the annulus between r and rz, then [ ° 2upr ar fr Pe x(ny? - 72) DyNACA TN No. 1795 87 or 22 2 y2 Pra Yana ie” - Ta") = Pye Va,2 (t2,47 - 22") a 2 ye es 21, 2 Pea Me Vee | Pa ae > Un? bry2 (Payee) %e,2 Paya (ree - 42,47) hence Prd q = 7 G2] fat (7) at/} bre “Pn, ‘The change of total enthalpy across the rotor is Ey - Hy = w(f, - £3) ‘Then Fey, te -ty) Hh Hh For compressor a,b ee 2 ey , Mab Vath Fan 82h” Yan y, y Tie Gye Vz,2,8 eH vhere *2,h yp "6,2,h ~ %0,2,b the quantity D748 to be assigned by the 21) designer, ‘Inasmuch as58 NACA TH No. 1795 = 2 3 z “ae Tye ¥, fue d 0,4, £) O-) D2 Taye 2 wed? wh (eat) , 4 Ton *) m2 Oye The where M, is the limiting Mach number to be chosen by the designer; 2h y a 3 9,1,h Fun Venn Fun 27272 7 Orty ry U; Bay ‘it Ut Ve,1,b Z 4 a e sy “Fe Q "pa 2, :) +4 Ge) (Vo,1,% (7-2)? Wat a4) Tht (c8) The pressure distribution at each station is obtained by rais- ing its density distribution (equation (C4)) to the power 7. The pressure changes between the stations at different radii are obtained by conbining these pressure distributions vith the pres- sure change across the rotor at the radius where the value of the polytropic exponent is known or assumed. The angle that the gas velocity makes with the axis of the machine at any radius is obtained from the known tangential and axial velocities.seot NAGA TH Wo. 1795 59 Symmetrical Velocity Diagram and Constant Total Enthalpy For the case of nontapered passage, r, =r2 =r and from equations (28) and (32) ab at, @°ae* ee ae 127 z (c9) 2 1,4, |. © at Le teSre 2 or (cio) Pe 7-2 0 wr Oe ee 22 TOT ee ee 2%: 4,2) 3 where tho minus sign is used for station 1 and the plus sign for station 2. Integration from r to r, yields 8, x 6 Lt t Et tog, or60 ; NACA TN No, 1795 2 0, fey \2 fy 7a 2(% *) ¢ 7) ¢ a +8, Fe ca - ( Ue) Vee) \? a where the plus eign is used for station 1 end the minus sign for station 2. The variation of axial velocity is obtained from equation (26a): (c1z) Integration from 7, to r gives y, f ¢ y 2 ‘) z z4b) i Th’ (_z \' -i(, £8, lo (013) Uj 2 @ 2) =P 18. Ge Vt ae Te where the plus sign in the last term is used for station 1 and the minus sign for station 2. Equation (C12) also gives ay, ay, 8, Uy 42 2,2 za _ 5+ Ma, V2 -%e,2 e+? (ca4) In the case of a tapered passage, the gas is assumed to flow in conical surfaces, which gives the value of rz as a function of xy. Bquations (28) and (52) give the distributions of { and Yq as shown by equations (D6) cna (D7), respectively, given in appendix D. The distributions of axial velocity and density at 1035seot NACA TN Wo. 1795 . 61 station 2 are the same as those given by equations (D8) and (15). After these distributions are known, the distribution of specific mass flow G is known and the radial displacement is found from equation (B5). In compressors of this design, the maximum value of 5 ° ayn Fin “9,2 7 Yer is usually at tho hub, Its value there is to be Vz sot by the designer. ‘Then 2,b eae cten tan Fin 2207 VM ya Tho h,¢ Vz1,h Ut r, 2h by, on -¥ y; War ran cc een oy = one = fb “pub a (c15) FL,t “it Sat z,1,b Tm this type of design, the limiting Mach number is usually at the hub. Hence the denominator of the last term of equation (C8) should be replaced by (222) Bagi n ~ Yo,1, sy ei C22) (gaa a (‘ea OF am ra 2 Wye (ez ‘The rest of the calculation is the same as in the previous design. Symmetrical Velocity Disgram and Condition on Axial Velocity For constant work over the blade height and nontapered passage, equations (C9) to (C11) still hold. If dVz,2/ar equals 0, equa- tion (27C) under Simplified Radiel-Equilibritm Celculation gives ay, Yea Sed = Bn 0 gor = SE Ze NACA TH No. 1795 ‘Then ie y, 2B, Uy 42 na, Bea ag 2 %z,1,¢ Ve ,ayt T ‘he enthalpy variation for this case is obtained by using equa- tions (26a), (27a), and (C9) Th Me iy, Mela (yx, Tate at) y, a a te ez Oe 2 2 _ et [Og eta ere) aed) ria Te . zi ae) + Bo, log, (c17) ay, On the other hana if eto 2 yy, 2 Ue 5B 2,2 ae 7 (c19) In order to make the distribution of H independent of 6,, the compromise condition ay, a2 ‘a1 B Ve,1 ~g + Va,2 ee = 0 1035,NACA TN No. 1795 oS may be used. When this relation is combined with equation (27c), there is obtained a2 Vv, Fea! gh n (r+ tobticg F (c20) ‘252,) Ve,1,t° °F, y, 8 12 CLUS (cea) 2,2, Ve,2,t" 7, When equation (26a) is applied to stations 1 and 2, and the two resulting equations are added FB (cza) If the requirement of constant work over the blade height is abandoned, it is possible to require no change in the axial- velocity distribution across the blade rows, which is the incom- pressible form of the condition of no radial displacement 1 V2 ay, Men, Me.2 25 29 (c23) ory ys When equation (23) 4s combined with equation (27) for the case of nontapered passage ty ah te te, (te th) 2 “2 & co (c24) x z @ ar, By substituting the symmetry conditions (equations (31a) and (31b)) into equation (C24)a ACA TN No. 1795 at. at. 3h ery -t) @- 8) v(om - 2%) Simplification yields (ces) or Ve eens (258) 6,1,¢ From equations (52) and (C25) ba 2 ¥ ae (_=_ (cas) ee” Ge) (c26a) eleieenyst t (ara VYoj2,t *,e \FL,t ‘Thus wheel-type rotation exists both in front of and behind the rotor. By equation (27), Sa _ ah (ee _ 82) nies dre Ore nar = 2 (ta,e - bat) TL,t (ca7) ~ 2,2 (SS) which means that the work done is proportional to the square of the radius.ACA TN No, 1795 APPENDIX D METHOD OF CALCULATION FOR ZERO AND INFINITE ASPECT RATIOS ‘The method of calculations is given for the two types of design used in the numerical examples. Free Vortex and Constant Total Enthalpy Zero aspect ratio. - In this type of design, the zero-aspect~ ratio case is the seme as that of the simplified-radial-equilibriun approximation. Infinite aspect ratio. - By equations (47) and (A17) 82781, i arr oa) 8)-8,\7-2 ¥ 2 2 2 2B - (Von + Vea) Ya ° moO, Pa () 2,2 By - (Vo,2” + Vz,2°) fn additions] relation between V,,1 and Vz,2 is necessary in order to solve the equation. In the section entitled "Infinite- Aspect-Ratio Calculation" two equations-ere suggested. For this design, equation (48) gives ea, Meg ar, "arg or Yy,a + Vay = constant (2)65 NACA TH No 1795 end equation (49). gives a av, Vga et + Ve 2 Gee = 0 o 2 2,1 + Vz,2° = constant (03) Also from equations (E4) ahd (85), neglecting the square term in Ay ¥z,1 Ya,2 = constant (04) ‘The three preceding equations give practically the same results. A convenient procedure of calculation is as follows: (1) In order to compare the result with other cases, the same value of ¥, 1, may be used. From equation (D1), Vz,2m 18 determined. (2) Insert these values in either equation (D2), (D5), or (D4) to obtain the constant in the equation. (5) Assume a number of values of Y;,13 cbtein V;. by the ano equation. Then use the following eqéation, which 13 obtained fron equations (D1) and (C3), to aolve for r/xy oe 2 y, get] (vp 2)? - (ve, 2)? 252, ” ” (4) Plot ¥z,7 and Yz,2 against x/rz, and obtain their values at the valies of x/rj desired.ACA TN Wo, 1795 67 When the distribution of axial velocity is lmown, tho density variation at any station is obtained by applying equation (A17) at the station qe 2 ( et ) a \eomane 5 t 2 2) 2 (2%) 1035 (05) @ @ ‘The pressure variation at each station is obtained by raising equa- tion (D5) to the 7 power. The pressure changes across the stage at different radii end the air angles are obtained in the same manner as in the simplified-radial-equilibrium calculation. Symmetrical Velocity Diagram Plus Constant Total Enthalpy Zero aspect ratio. - With radiel displacement not equal to zero, the equations for tengential velocities are different from ‘the expressions of equation (C10). From equations (28) and (32) (Ds) 5 5 Ome Tunes oe ceeeeiteee end Yea. $2 Re nts (o7) ¥, x, eye eeece a) Tye Fe ®68 NACA TW Wo, 1795 Fron equation (26) with the last term negligible for this case, inasmich a8 rg - 7) 1s much less than L, ave, at pi eee ced est ary =~ 7,8 fs Oey For station 1, from equation (D6) : a afr 2s ake Bee Integration fron rm, to rv, yiclds 2 2 2 2 Yet) _(Test2b) 5 tog, ZL 2 (20 Tum The v. 4 er 2 z Tit Tit Which is the same as equation (013). For station 2, 2 ay, ry By U; dr; ey mr, Tb Ot Uae 2 %2,2 “ey = (CE 2 z On r L (te? Va,2 AV, 2 ao arg" rye + By Ue ) ary . x? 2a (71 be 6 TL :) 2r)@ vt [Fe ” Fr P14, 71,6 Integretion from 11, to 7 yields a T1,t 2 4 fe . 2)” _ (Yeas) _ a(Z2) Us "1,4, TLh Tit which differs from equation (C13). 10351035 TACA TH Wo. 1795 69 ‘The density distribution at station 1 is the same as the simplified redial approximation, whereas that at station 2 is obtained by using equation (D5). The solution of this case is a Process of successive approximations. Values of rg(rj) obtained in the simplified redial approximation can be used here as the starting values. Then the distributions of Vg,2, Vz,2, 2s and are ‘caloulated from the preceding equations, and new values of rp(r,) are computed from equation (B5). Usually, only two or ‘three cycles are necessary to obtain the correct value, as the difference between this case and the simplified-redial-equilibriu approximation of this type of design is emall. Infinite aspect ratio. - The first equation for the condition @, = Gg 4s the sano as equation (D1). The second equation necessary to solve this case 1s a little more complicated then that in the previous type of design because at #0. If equation (49a) te used tea BE +2 Bt 3G BB) =< 2 (hh + fp) = -0% ‘Thon y, ng + %,,2° = - 0? + constant oe Teal Tad an uy?” UL? In onder to compare the result of this case with other cases, the sano value of Vz,),_ may de used. Then from equation (D1), Yz,24m 18 found, and the omstant’in equation (D10) is evaluated by uding this set of Vz,1,m and Vz,2,_- A few values of Vz,1 are assumed at any other given redius, with corresponding values of Vz, obtained from equation (D10}. ‘The correct values of Ya, and Vz,2 that will satiety equation (D1) are obtained by interpolation:70 NACA TW No. 1795 Aftor tho aistribution of axial velocity is Imovn, the density Aisteibutions are obtained from equations (C11) and (DS), and the pressure distributions, total enthalpy change, and air angles aro obtained in the same manner as before. seotNACA TW No. 1795 n APPENDIX E APPROXIMATE VALUE OF RADIAL DISPLACEMENT ACROSS BLADE ROW FOR GENERAL CASE IN WHICH Hy, He, £1, AND fp ARE DETERAINED BY DESIGN AS FUNCTIONS OF ry In the latter part of appendix D, distribution of axial veloc- ity 1s expressed in terms of mon H, f, ry, and ro(rj). Alternatively, this distribution can be expressed in terms of radial dieplacement and its value determined by the simplified-radial- equilibrium calculation, for which 4 = 0. For the case of nontapered passage, it is seen from equation (21) av, 1 See 18 2 Ye. Gy 7 %2,2,8 art 2 (8) y, re (m) ay, ay, at, e.2 22:8 40 (x ate Yen? “ary "2,28 an (sy (ree) = = thom b0\-2 By substituting (r+ 49) for rp, expending é + 3) ane nomial aaa be binomial series, neglecting terms of greater order than =) , equation (£2) becomes ay, 232 %2,2 Gr, * %z,2,8 at, be ake (50,2 (9-68)| : 1 If ty (24) 42 know, equations (#1) and (85) may bo solved as linear first-order differential equations in ¥,,1° and ¥,,2", respectively, giving (omitting the subscript 1 on r) we OL fo { (& Bi e (zs) and ce 24, 39)? #) ote) 2 2 ea 2 'e. a 21 \t, 2 Teo" = 220 +3 abe ae & te |e or + Ve 2m (Bs) vhere er) = [ end subscript’ m may here refer to eny radius between hub end tip. S6LT "OM AE VOVEENACA TH No. 1795 i For the Liniting case of zero aspect ratio, the last tem in equation (E1) approaches zero, so Vz,1 = Vz,1,91 whereas Vz,2 is obtained by integrating equation (£5) with the third term neglected. 2 2 2 2 Va,2" = Ve,2,87 - (a,2,0jm° ~ Ya,2,m°) + zh , ‘m (Bs) ‘When equation (£4) is integrated by parts and 4, (r)) is replaced by Yq g(r) as in equation (51), there 1s obtained in which4 ACA TH No. 1795 If it 1s desired to compare the general case with other cases on the dasis of the same Vzia.m) then V1.7 Vz,1,n,0° By the use of the mean value theorem Of integral calculus, the preceding equation gan be written as 2 Ye ® (F) %e,1 > -4 =V,1,8 (z7) where Vz,1,5° 16 @ moan valuo of Ye,r,0° between rand yy the mean depending on the choice of the function g (r). if the approximation te made in letting Vz,2,. = Vz,2,m,q» eauation (BS) may be written as Ye %2 (r) ene) oP 4 (x) (za) where ve) Boo, 0 +B, {fe oi} - [2 6u*} 3 (r) <6 fe) a ey ° er @& 9) sey a ioe : 7 end {2 4 (r) 6 @) : ef ° %; m 2 z (r) (f) 9 Lal? $61035 ‘The change in the distributions of V,,; and ¥,,2 with the marimm dieplacament y, fora g given _g (r) is now determined by differentiating equations (B7) end (B8), aseuming that r,, 4 Vz,1,0-» and Vz ,° are infopendent of y, = 5 i) a Jo Pz (r) . 2 hs 2069 Wo,2 = = EE 7 4 (2) +7 { [ate]? = [etal] SBE), to ol 1) ono) Yn2° Ye,2 We y, 2 2 (2) - [alta @3(r) sry HOME = tea] 4) uy 3 Oo Yu,2 a76 ACA TN Wo. 1795 ‘Tho equation of continuity, equation (14), may be written as Vz. yeelr) Pulz,1 x PaYetaa ino aye & «| (giz) by replacing rg with r1 +o g(r). Tt 1s here being assumed that the displacement a," for the continuity equation hae the sano fora aa, but uay differ in magnitude fron, Jy. If tho variation of py/og (r) with y, 1s neglected, that ie) tho density aistribu- tion te aseuned to be deterained prinarily by the tangential veloc- Sty distribution, differentiation with reapest to gro for a given g(r) gives : 5 Se dat) ah 106, 4 2 (m3) fo Mn,2 ayy ae) dive & alr) If the came distribution of V,, and V,2 satisfies both the continuity and equilibrium equations, yo is a function of yo determined by the differential equation, which 1e obtained by dividing equation (E11) by equation (£13): 2 ae @) Py () ve? lee] r + nye EL" hy, E Cele) 2 10) Tad EF me ! ; Sats" Ve,2" [ee]? = Le(mgi]? | @5(r) on L+% a %() Va,2° (s) ©, (r) seot1035 NAGA TH No. 1795 7 In order to evaluate equation (B14), the form of g(r), which is implicit in the equations used, mst be found. An order-of-magnitude result may be obtained, however, by equating the right-hand side of equation (E14) to a constant -K* and ‘determining the value of ‘the constant from the boundary conditions on g(r). Because this assumption involves setting dy,/dyg equal to a constant, it is equivalent to the ascumption already stated that for the selected 4o = Yo B(x), ‘the corresponding 4, differs only in amplitude. In order to obtain the order-of-magnitude result, the right-hand side of equation (B14) is simplified by (a) Setting the first two terms in the bracket equal to 2 (b) Considering the terms involving y, end yg nogligible when they are compared with unity (c) Ignoring the last term in the bracket because it contains L, yeitten in tems of r/r, instead of r, the term 2 ¥e,2° () im the denominator. (If equation (B14) 1s 2 2 (5 decomes Gy » Which 1s about 250 for A =2 r and = = 0.6.) * As a result of this simplification, equation (E14) becomes ft) 91 (©) = alr) (zis) Rewriting equation (E15) gives = «gy Ba ©) AGL s Bate) (zs) ‘When equation (E16) is differentiated with respect to r and the relation $9, (r) = g(r) 10 used, 2 2 gol) +b a ete) + le -4 a(z) = 018 NACA TN No, 1795 This equation gives g(r) as a Bessel function of the first order end argment (x/IK). The value of (x/K), end tms of K, ie 7 determined by the boundary conditions g(r,) = g(ry) = 0. In onder that g(r) have a single maximm, the first eigenvalue of this Doundary-value problem must be taken. A satisfactory approximation to this solution may be obtained without involving Bessel functions by replacing g(r)/r in equation (E16) by g(r)/r,, differentiating, and solving: ssot z ele) = 6 (x, cos ir + Ky ein Ir) - NG) 3-ape ‘The boundary conditions determine, using the first eigenvalue for K, x (B17) and therefore, (4-4 oe DY we (rpm arg? (yam)? (This approximate equality is correct within 1 percent for (rp/ry) > 0.5.) Substituting this result in equation (E15) gives 8 ay, ” Bee-x-- meh (ei) ‘In this very rough approximation dy,/ay, is therefore equal to minus the square of the aspect ratio. By integrating equation (£18)NACA TH.No. 1795 79 and by letting yo,5 equal the value of y, corresponding to Yo = 0 (sinplified-redial~oquilibriu approxination) : Jo = Toe ~ 4? Yo (m1) ‘A solution corresponds to ye = Ye = ¥, which when substituted into equation (B18) gives 2 a8 +a? b= as, (so) Raden, P.: Investigation of Single Stage Axial Fans. NACA TH No. 1062, 1944, 2. Howell, A. R.t Fluid Dynemios of Axial Compressors. Wer Emergency Issue No. 12 pub. by Inst. Mech. Eng. (London), 1945. (Reprinted in U.S. by A.S.M.E., Jan. 1947, pp. 441-452.) Eokert, and Korbacher: The Flow through Arial Turbine Steges of ‘Large Radial Blade Length. NACA TM No, 1118, 1947, 4, Lamb, Horece: Hydrodynamics. Cambridge Univ. Press, 6th ed., 1932, articles 10, 529, 358. 5. Vezeonyi, Andrew: On Rotational Gas Flows. Quarterly Appl. Math., vol. 3, no. 1, April 1945, pp. 29-37. 6. Sinnette, John T., Jr., Schey, Oscar W., and King, J. Austin: Performance of NACA Eight-Stage Axiel-Flow Compressor Designed on the Basis of Airfoil Theory. NACA Rep. No. 758, 1944. 7. Pochobradsky, B.: Effect of Centrifugal Force in Axial-Flow Tarbines. Enginsoring, vol. 163, no. 4234, March 21, 1947, pp. 205-207.20 NACA TN No. 1795 (a) Stream surface over four stages of multistage turbomachine. Vp 1) + Yo J Nabe Cire \e F — BND TUT UTUTT ETT (b) Intersection of (ed Intersection of stream m surface with surface with axial plan plane normal to axis. Figure 1, - Stream surface over four stages of multistage turbomachine and its Intersection with planes normal to and containing axis of machine.NACA TN No. 1795 8 2 kil il It 7 1 ! | I | a a ele [rs re [omer ieee * (a) Wontapered passage. 1203 bi | | 1 L i} fl "qi L Th 4 rio fra |e ta - “2 to Ne Tapered passage Figure 2, - Stations between blade rows.NACA TH No. 1795 Torque SEE Figure 3. - Stations i and j short distance apart.1035 NACA TN No. 1795 Grey = 12 ry Te > th 83 he +03 102 01 (a) Distribution of specific mass flow. yy (b) Radial displacement across rotor. Figure 4. - Synmetrical-velocity-diagram and constant-total~ enthalpy compressor.nlite “9 ae ye 2, approninate Figure 4. - Continued. Vz/Uy, (cl Variation of axial velocities. Synmetrical-velocity-diagram and constant-total-enthal py compressor. ssot +” G6LI ‘ON KL VOWN *1035 NACA TN No. 1795 Yo,2/Ut ey, ——A's0 anna? —-—A+ am sereee ——--A = 2, approximate ws Soa EF 7 GLE ° i 8 2 T.0 rire (4) Variation of tangential velocities. Figure 4, ~ Continued. total-enthalpy compressor. 85 Symmetrical—velocity-diagram and constantNACA TN No. 1795 60 Aso —- Ae? ke a seree. ———A-= 2, approximate EE 20 O 7 zi 78 9 T.0 rire (el Variation of air angles. Figure 4. - Continued. Synmetrical-velocity-diagram and constant- total-enthalpy compressor. ssotNACA TN No. 1795 a7 8 -— }-—~ a sur = A+ Oors.ree 7 +2 = 0 =~ A= 2, approximate 6 7 8 9 1.0 rite (f) Variation of Mach number relative to rotor blades. Figure 4. - Continued. Symmetrical-velocity-diagram and constant~ total-enthalpy compressor.88 NACA TN No. 1795 ai er Aez tat Ate =n A+ 2, approxinate toh (or s.s0e {— += 7 78 2 T.0 te (a) Pressure distrivutions and pressure rise across rotor. Figure 4, - Continued. Synnetrical-velocity-diagram and constant= inthal py compressor.| { | i Figure 4. - Concluded. J. a at tip Ue At moan radius ay At hub (h) Velocity diagrams at different radii. =e Synmetrical-velocity-diagran and constant-total-enthalpy compressor. SON NL VOYN S61 6890 NACA TN No. 1795 a -90 fa) Distribution of specific mass flow. .o12 | .008 an ale 004 / ae aa: : 7 7 3 7,0 cine =a (b) Radial displacement across rotor. Figure 5. - Free-vortex and constant-total~ enthalpy compressor.D 1035 8 A= 0 or s.r.e. A 22 4 16 Vz/Ut (c) Variation of axial velocities. Figure 6. - Continued. Free-vortex and constant-total-enthalpy compressor. G6L1 *ON NL VONNACA TN No. 1795 92 50 = 8 4 « 40 7 4 30 20 A= 0 or s.ree. - Aso to 6 7 <8 “9 T.0 rire (4) Variation of air angles. Figure 6. - Continued. Free-vortex and constant-total-enthal py compressor. seo1035 NACA TN No. 1795 8 7 6 A+ 0 or sere. SES —-— A+ 5 6 7) 8 1.0 rire (e) Variation of Mach number relative to rotor blades. Figure 5, - Continued. Free-vortex and constant-total-enthal py compressor. 93NACA TN No. 1795 94, 1.05 a = 4 a 1.00] > a & A+ 0 or sree. 95 r SEE KEO 4 90 1.05; Sar -98 7 rile Pressure distributions and pressure rise across rotor. w igure S. - Continued, Free-vortex and constant-total-enthal py compressor. ssoTAt mean radius At hub (9) Velocity diagrams at different radii. Figure 8. - Concluded. Se Free-vortex and constant-total-enthel py compressor. 6LI "ON NL VON 66= Teh NACA TN No. 1795 to |-A+0 of s.r.e., behind rotor ‘A= 0, in front of and behing rotor Lo vot | oe sg _| 2 et 8 {al Distribution of specific mass flow. 2025 [—] 020 2015: “4 ——— A+ Oorssr.e. —-— Aw 005) / 7 “8 rly (vo) Radial placement across rotor. Figure 6. - Free-vortex and constant-total-enthalpy turbine.' 1035 O or s.ree. 8 fe) Variation of axial velocity In front of rotor. Figure 6. - Continued, Free-vortex and constant-total-enthal py turbin T.0 "ON NL VON g6L198 NACA TN No. 1795 $8 45 X 2 Uy 2s «, deg Is . 7 6 2 T.0 eye (a) Vartation of gas anglé entering rotor blade. Figure 6, - Continued. Free-vortex and constant-total-enthal py turbine. sgot1035 99 NACA TN No. 1795 1.0 e . Zs . a 6 ~ => 8 BS | z. == id A= 0 or ser. >—-S =. 3 7 3 $ ne Variation of’Mach number in front of rotor. fe) Free-vortex and constant-total-enthal py Figure 6. - Continued.NACA TN No. 1795 100 \ seot / A+ 0 or sree. i —-— fee 8 1.0 rire (f) Pressure distribution. Figure 6. - Continued. Free-vortex and constant- total-enthalpy turbine.1035 A= 0 or s.teee At mean radius At hup (g) Velocity diagrams at different radii. Figure 6, ~ Concluded, Free-vortex and constant-total-enthalpy turbine, TON NL YOWN a 8