US2943839A - Elastic fluid mechanism - Google Patents

Description

y 1960 R. BIRMANN 2,943,839

' ELASTIC FLUID MECHANISM Filed May 10, 1954 8 Sheets-Sheet 1 FIG.

FIG.

INVENTOR. UDOLPH BIRMANN ATTORN-EYi dwlam July 5, 1960 Filed May 10, 1954 R. BIRMANN ELASTIC FLUID MECHANISM FIG. 3.

FIG. 2.

8 Sheets-Sheet 2 FIG. l4.

INVENTOR. RUDOLPH BIRMANN ATTORNEYS R. BIRMANN ELASTIC FLUID MECHANISM July 5, 1960 8 Sheets-Sheet 5 Filed May 10, 1954 INVENTOR.

RUDOLPH BRMANN BY flan, La 1 4 ATTORNEYS y 5, 1960 R. BIRMANN 2,943,839

- ELASTIC FLUID MECHANISM Filed May 10, 1954 8 Sheets-Sheet 4 INVENTOR. RUDOLPH BIRMANN ATTOR N EYS y 5, 1960 R. BIRMANN 2,943,839

ELASTIC FLUID MECHANISM Filed May 10, 1954 8 Sheets-Sheet 5 FIG. IOA. FIG. IOB

INVENTOR. RUDOLPH BIRMANN ATTORNEYS July 5, 1960 Filed May 10, 1954 8 Sheets-Sheet 6 RUDOLPH 'BIRMANN ATTORNEYS y 1950 R. BIRMANN 2,943,839

ELASTIC FLUID MECHANISM Filed May 10, 1954 8 Sheets-Sheet 7 INVENTOR. RUDOLPH BIRMANN FIG. l9.

ATTOR NEYS July 5, 1960 R. BIRMANN 2,943,839

ELASTIC FLUID MECHANISM Filed May 10, 1954 8 Sheets-Sheet 8 FIG 20 RUDOLPH BIRMANN l K INVENTOR. l I

ATTORNEYS 2,943,839 ELASTIC FLUID MECHANISM Rudolph Birmann, Newtown,-Pa., assignor to De Laval Steam Turbine Company, Trenton, Nil, a corporation of New Jersey Filed May 10, 1954, Ser. No. 428,627

20 Claims. (CL 253-39) This invention relates to elastic fluid mechanism and has particular reference to improved designs of impellers, diffusers andturbines, the invention being, in part, directed to the design of the bounding surfaces of the elastic fluid passages of such devices; The inventionrelates also to an improved type of gas turbine.

This application is in part a continuation of my application, Serial Number 38,995, 'fi'led July 16, 1948, which was in part a continuation of my application, Serial Number 614,791, filed September 6, 1945; now Patent 2,623,357, datedDecembrBl), 1952.

One object of the present invention is the embodiment of novel principles of design in flow passages for elastic fluid in which changes of its thermodynamic state are being effected. In accordance with this object there is provided a novel type of compressor embodying one or more impellers and associated diffusers. Also within the scope of this object is the provision of a novel type of turbine, the same basic principles applying to all of these fluid-handling elements. Basically, this aspect of the invention comprises the attainment of designs to satisfy the requirements of high efficiency, the distribution of work input and output to avoid aerodynamic overloading while properly taking into account the laws governing stability of flow, and the securing of other conditions to provide maximum efliciency and to avoid losses. In brief, these ends are accomplished by the design of either rotating or stationary passages in conformity with the laws required for stable, threedimensional, flow, i.e., flow which fulfills the differential equations for the flow of an elastic fluid without'separation from its boundaries;

In design in accordance with the invention for the attainment of these, objectives it is found that, in the case of rotary devices such as compressors and turbines, the blades, if made in the former single continuous surfaces extending from entrance to exit of the elastic fluid, will generally deviatevery substantially from radial condition and hence would be subjected :to enormous centrifugal stresses in high speed devices to which the invention is particularly applicable. Accordingly, a further object of the invention is to provide'blading for such devices consistent with. the. fundamental desired principles but having a construction such that centrifugal stresses are kept within'properbounds. This object is attained by either providing rows of blades. which do not deviate to a non-permissibledegree-from;radial condition but, in fact, serve to support each other to provide a blade structure of. adequate rigidity and strength, or by adopting for the blade surfaces (in the case ofsingle blades) a three-dimensional-type surface which is characterized by possessing atleast one, and preferably notlessthan two, radial anchoring lines which impart sufiicient strength to the blades so that non-radial portions of the blades re-' turbines involving improved hub-construction so thab light weight may be achieved consistently with high me- 52,943,839 Patented Jul 5, 1960 chanical strength. Hubs in accordance with the invention are in the form of convex shells, the stresses in which are, at any point, substantially only in tension with the'res'ult In the im that light weight construction is secured. proved construction thermal stresses are minimized-as compared with the conventional discs.

The invention also provides a method of producing blades which are designed entirely from the point of view of the aerodynamic principles embodied in the invention, and are not restricted due to considerations of manufacturing, and it provides a novel method of producing-such blades and fastening them to the rotor hubs.

The invention also serves to provide very desirable small discharge angles. As is well known incompressor design, small discharge angles result in high efliciency and desirable characteristic curves of performance.

However, such discharge angles have heretofore not been attainable without involving design difiiculties. With small discharge angles a higher degree of reaction is obtained, or, in other words, a much greater percentage of the total pressure rise is produced in an impeller as a direct result of centrifugal effect, which is a highly efiicient process, and a smaller percentage of the total pressure rise must be obtained in a relatively inefficient diffuser.

In accordance with the invention, substantially more than 50%' of the total static pressure rise is obtained in the impellers, the remainder being secured in the diffusers, which, in accordance with the invention are made unusually efficient, this high diffuser the present invention that the flow leaving the impeller and entering the dilfuser is unusually homogeneous, that is, at equal distances from the center of rotation there is little or no difierence in the direction and magnitude of the velocity components, resulting in minimizing the socalled mixing losses arising from momentum transfer.

A further object of the invention is a provision of a novel type of gas turbine in which, among other features,

the compressor blading and turbine blading are carried by a common hub with various attendant advantages which will hereafter be set forth.

These and other objects of the invention particularly relating to details of construction and operation will become apparent from the following description read in coniunction with the accompanying drawings in which: Figure 1 illustrates in axial section the assembly of the parts of a five stage compressor constructed in ac- Figure 6 is a perspective diagram illustrative of certain geometrical elements involved in the design of an impeller;

Figure 7 is an axial view of the matter illustrated in Figure 6;

Figure 8 is a diagram illustrating a meridional projection of Figure 6, with an auxiliary diagram at the upper left portion of this figure looking in the direction of the arrow to show the true value of the angle y;

Figure 9 is a diagram explanatory of certain of the steps in design procedure hereafter set forth;

Figures 10A and 10B are diagrams illustrating certain considerations involved in two types of approach flow to an impeller;

efficiency being achieved by so designing the impellers in accordance with various dilfu'sers.

concentricity with the rotor by a rigid structure 14', the

design principles; f V

-Fig'ure l6 is ,a diagram illustrating certain dimensions involved in an expression given for vanes at (7) in Figare 15;

Figure, 17 is a diagram explanatory of another fashion in which'vanes or blades provided in accordance with the present invention may be defined;

. Figure 18 is an elevation, partly in axial section, illustrating a gas turbine constructed in accordance with the 7 invention, the turbine being particularly designed for constant speeddrive;

Figure 19 is an'enlarged section on the plane indicated at 19 ;19 in Figure 18 showing the hollow construction of -a-turbine blade; and a a l 7 -Figure 20 is an elevation, partly in axial" section, illus- .trating another form of gas-turbine constructed in accordance with the invention and particularly designed for variable speed drive.

' .In order to provide an understanding of the'invention the description will first be appliedto the, design of a five stage compressor illustrated in assembly in Figure 1. A" rotor is built up of a series ofindividual rotor elements,

one for each stage, designated generally as ,2, 4, 6, 8 and rod 12 as indicated. The air passages through the stages have their circumferential outer limits defined by a hous- 10. These are held together axially by means of an axial to the design of axial flow impellers jn'which' the mean.

ing 14 with which the'blading of the impellers has suitable small clearances and which'supports the vanes of the Thehousing 14-.is held in proper housing 14 being free to slide axially within this strucat 24. The various associated diffusers'have' their vanes indicated generally at 26, 28, 30, 32 and 34. These diffuser vanes are secured to the housing 14. In advance of the first stage impeller there are provided directing vanes indicated at 36 so arranged as to impose free vortex flow (or other desired flow as hereafter pointed out) on, the air entering the first stage impeller. These'vanes are conventional and are of the type used to provide free vortex flow in axial flow compressors.

Bearings 38 and ,40 mount the rotor of the compressor.

Referring to Figures 2 and 3 which show the impeller 4 of the second stage, the hub comprises a convex portion 42 in the form of a body of revolution supported by discs 44 46 and providedwith a cylindrical strut 'or com-.

pression member 48.- The fore and aft extensions 50 and 54 of the hub, are toothed as indicated at 52 and 56 for interlocking with the preceding and succeeding impellers.

As will be noted from Figure 3 the hub' is made in two sections for purposes'of simplicity of construction. These 'two's ections are brazed together to form the complete hub. As will be pointed out in greater'detail hereafter the'hub is made in the form illustrated so that the periph- 7, 'eral portions thereof when loaded by the vanes are substantially only in tension'to secure a light weight but In Figures Z'a'nd 3 each of the, vanes consists of a single continuous section from the inlet edge to the outlet edge.

; 'In some cases'fas' will be more fully'pointed .out here after, the'vanes are desirably made npoftwoor more "ture. The blading of the first stage impeller is indicated 1 generally at 16, that of the second stage impeller at 18,

7 that of'the third'stage impeller'at 20, that of the fourth 1 stage impeller at 22, and that of the'fifth stage impeller I sections and for the purpose of illustrating this Figures 4 and 5 show another impeller in which this construction is adopted. The impeller generally indicated at 62 has a convex'portion 64 carried by the discs 66 and 68 which are joined by the cylindrical compression strut 70. .Ex-

tensions 72 and 74 are provided .with teeth 76 and 78 for interlocking with adjacent impellers as in the case of the modification previously described. Each of the vanes of this impeller comprises two sections 80 and -81which as indicatedin Figure 4 overlap each other and may be Q connected together for mutual. support where they meet; the connection'being-efiected,for example, by welding. Alternatively, the vane sections may merely support each f "other by leaning I against each other, this resulting in excellent vibration damping, The inlet and outlet edges of the first'section 80iare indicated at 83 and 85 while the inlet and outlet edges of the'second section 81 are indicated e at 87 and 89-respectively. In Figure 5 various radial planes are designated by Roman numeralsand theinter;

cepts of these planes'with one of the vanes are indicated in Figure 4 by corresponding Roman numerals.

' Theoretical considerations show that the maximum ef- V ficiency of the impeller passages will result when these passages .are of the mixed .flow type, i.e., when inflthe progress from inlet to-outlet the flow of the elastic fluid;

being handled has a suitable radially outward 'component of motion with discharge at an average radius greater than the average radius at the inlet. This is in contrast:

radius of flow is substantially constant. It is also in contrast with ordinary centrifugal "impellers in which the axial length of path is a minimumconsistent with smooth- 1 ly changing the directionof flow from an axial approach 'to a discharge with substantially no axial component of motion. (However, vvarious aspects of the invention are applicable to impellers ofboth of the last mentionedtypes to improve their efiiciencies.) g V Y Referring now to Figures 6-, 57; and 8, the impeller passages are'interiorlybounded by a surface of revolution indicated at 84 and constituted by the hub surface, this surface of revolution, in the case of the. present second stage impellerv to which these figures relate, extending from a substantially cylindrical portion at the inlet'to a substantially cylindrical portion at the outlet with smooth variation of curvature betweenthe inlet and the :outlet, l

the-latter ofrwhich is at aisubstantially greater radius than the inlet. It may be here noted that the fact that 1 this surface of revolution becomes'substantially cylindrical at both inlet and outlet is not of. major significance in the present design and, as will be pointed out hereafter, this the fifth stage.

ability when the inlet is from an approach flow, which does not have a substantial radial component of motion and when the outlet is arranged for delivery to a succeeding stage requiring a radially inward deflection of flow.- Since work on the, elastic fluid inian impeller should 'be' accomplished when the flow has no radially inward component of motion, ,the'cylindrical inlet and outlet condi- 1 tions merely represent limits; at -both'inlet and outlet the surface of revolution such assflcould'be tangent to cones erally similar inits governing conditions "toithe surface 84. This surface 86-follows ageneralway the surface" 7 n 8'4, generally-having greater radial spacing from tlie-lat F e terj surface at Sits'inlet than at its outlet; The surface 86 ing ofthe impeller has slight clearance.

isdefined by the inside of the casing with which thebIad- V J diverging in the direction -of axial pro'gress'of the elastic 7 The external circumferential boundary of the passages i 1s defined by a surface of revolution indicated at 86 gen-" erned by considerations of preceding and succeeding devices, efliciency (for example in securing a maximum percentage of pressure gain in an impeller due to centrifugal eifect), cross-sectional area sufficient to handle the desired flow, smooth turning of flow, Mach number, mechanical strength of the rotor, etc. As will become evident, hereafter, these boundaries are treated as arbitrarily assumed in the approach to the design of the blad ing in the cases of compressors, diffusers and turbines.

By virtue of the impeller blading the elastic fluid is caused to have between the surfaces of revolution 84 and 86 paths of flow relative to the impeller along stream lines, one of which is indicated in Figures 6, 7 and 8 at S. It may be assumed, with small error, that the flow is axially symmetrical so that any stream line, if rotated about the axis of revolution of the impeller, would, in any position, coincide with another stream line. Accordingly, if a stream line were rotated about the axis the locus of its successive positions would be a surface of revolution such as indicated at 88, any point of which would be on a particular stream line which would lie wholly in this surface. The surfaces such as 88 will be designated hereafter as stream surfaces. Considering a particular point P on the stream line S illustrated in the figures under consideration, an axial plane through the point P will intersect the stream surface 88' in a line M 'Which will be called a meridian line of the stream surface.

As will be hereafter apparent the variations of an angle 7 along any meridian line is definitive of all of the similar stream lines on any corresponding stream surface,.and consequently of the shapes of the vanes extending along the stream lines, and this angle will be clear from Figures 6 and 8 in which it is illustrated as the angle between the tangents at any point P to the stream line and to the meridian line of the stream surface passing through that point. The upper left hand portion of Figure 8 illustrates the projection which gives rise to a true value of the angle 'y. The angle, it will be noted, lies in a plane which is tangent to the stream surface at the point P. Before departing from consideration of these figures there may also be noted, as indicated on Figure 8, the angle x which is the angle between the tangent plane just mentioned and the axis of rotation. It is, of course, the angle of slope of a meridian line of a stream surface with reference to the axis of rotation. As used hereafter, the Z axis is taken along the axis of rotation, being positive in the direction of the flow. The origin of reference is taken at O, arbitrarily, as used heerafter, at the axial position of the inlet face of the impeller. The angle indicated in Figure 7 is the total angle about the axis of rotation subtended by a stream line from its inlet end to its outlet end. Figure 7, is a cordinate angle corresponding to any point such as P. Its origin is arbitrary, as will more fully appear in its use hereafter.

As applied to an impeller, the invention involves the determination first, of the stream line S which should exist between inlet and outlet of an impeller for the proper transfer of energy to the flow for assumed conditions of operation and for a choice, arbitrary to some extent as indicated above, of the inner and outer circumferential boundaries of the flow passages. Secondly, there are to be determined the vanes which will actually impose flow along these stream lines and which, for all practical purposes, must lie along the stream lines at any point, i.e., every element of area of a vane must, subject to consideration of vane thickness and proper fairing of inlet and outlet, consist of a surface which is the locus of a group of adjacent stream lines.

It may be noted that the curvature of the inner and outer circumferential boundaries will be taken into account in determining the vane shape; and also taken into account is the nature of the approach flow to the vanes; for example, in some cases the flow to an impeller, as in the case ofsome first stage impellers, may have no radial component'of motion, as indicated for example in Figure 6 10A,whereas in the case of succeeding stages, or in the case of the first stage illustrated in Figure 1, the flow has a curvature in its meridional projection into an axial plane, as indicated in Figure 1013. The resulting blade angles for these two cases differ substantially.

Involved in the first step is the satisfaction of the well known three-dimensional laws of flow and thermodynamics of elastic fluids, is, given the other boundary conditions the stream lines must be determined as proper solutions of the differential equations expressing these laws. In the present exposition, there will be omitted the lengthy and intricate'proofs that the succession of steps involved in the present design actually satisfies these laws; instead there will be set forth those steps involved in determining the stream lines which actually represent, in the form of a method of successive approximation, the procedure for solution of the various equations mathemati. cally stating the laws.

Various types of flow have been investigated, and it has been found that vortex flow proves to be especially advantageous for the described type of compressor: for a given limiting Mach number the twist of the blades along anyradius can be kept smaller than for any other flow. it happens that vortex flow has the advantage of allowing a relatively easy calculation compared with other types of flow. The following calculation procedure 18 based on vortex flow; i.e., along orthogonal lines there is assumed to exist a constant product c -r, a constant total energy and a constant entropy (0 being the whirl component and r being the radius). The foregoing is to be taken as the definition of vortex fiow as that term is used herein.

In cases where the Mach numbers are not considered as excessively high, other flow patterns may be applied, favoring small or zero blade twist, regularly distributed mass flow over the channel Width, and/or regularly distributed degree of reaction, etc.

Assuming that hub and tip contours have been chosen in accordance with the principles given above, which contours as will now be evident define surfaces of revolution 84 and 8d bounding the elastic fluid passages, with the meridian lines of at least one of said surfaces deviating substantially from parallelism with the axis, the design then proceeds with the assumption of certain constants as follows:

The vortex strength at the inlet, (c -0, c being the whirl component and r the radius at any point of the inlet.

The vortex strength at the inlet, (c -0 c being the same quantitities are evaluated at the outlet.

The angular velocity, u, of the impeller. (This is zero in the case of stationary passages.)

(Considerations such as the desired Work input per stage, permissible tip speed, etc., together with such a choice of (c m), as will result in low Mach numbers, small twists, etc. will determine these constants.)

The inlet temperature and the inlet pressure.

The weight flow, G.

The constriction coefficient 6.

The adiabatic efiiciency of the impeller (or difiuser). This may be taken, for original calculation, as thatof known impellers of similar capacity.

The design then involves steps as follows, reference being made to the various equations given in Figure 15.

Meridians, which may be defined as the intersections of stream surfaces with axial planes, e.g. M of Figure 6, are tentatively assumed as at 91, 93 and 5, Figure 9. The intersections of an axial plane with the surfaces 84 and 86 provide fixed inner hub and outer tip meridians, as indicated. it is convenient to choose these on an assumption of'division into four'equal parts of the mass flow through the impeller, and they are so indicated in this figure. (it may be noted that this choice is rather arbitrary in view of the adoption of the method of sucsidered, with the same sense of sign;

cessive approximations which leads ultimately to determination ofactual meridians.) I

fTentative orthogonals to the meridians are plotted at those locations where itis desired to determine the proper blade angles. In Figure 9 such tentative'orthogonals are indicated at 90,- 92, 94, 96, 98 and 1130, those at 99 and V .100 corresponding to the inlet and outlet ends of a blade. (When the term orthogonal isused herein it will be understood. to refer to anorthogonal'to the meridians in an axial plane.) J Based: on these "orthogonals, the

different steps. which-will new be given lead 'to the determination of new tentative meridians. .;As will be hereafter pointedHout, these tentativezmeridians. are then used ,for the drawing of a new set of orthogonals'and the procedure is repeateduntil convergency of the process leads to a sufliciently accurate approximation The meridians of the hub and tip contours being known and fixed, their curvatures are determined either graphically or analytically. Let; 7 v

k =the curvature'of the meridian of the hub-contour Such a curve is obtainable fromthe usual air tables. A curve of specific volume v against the exp'ressionj -J i c i 1 is now plotted for the outlet conditions assuming ait-eni- V ciency for the impeller derived from experience, e.gi 90%", and assuming .that throughout the extent of the' outlet the conditions he on an adiabatic'line. These two curves. j

ditfer so little from each other that interpolation for points other than at the inlet and outlet can be carried out with' sufiicient accuracy. i l a i The value of c at any particular point must be deter- V mined. Forthis, c for the orthogonal through the .pointunder consideration is assumed on a trial basis.

Then Equation 1 gives c for the point. The whirl com ponent c is now determined using Equation 2 in which at the'base ofthe tentative orthogonal being considered;

taken positive :for 'a -portion of the meridian concave towardsthe axisof rotation;

i 'k =the'cnrvature of the meridianlof the tip contour at the outer end ofthe tentative orthogonal being cons=thedistance from the hub to the point being consideredmeasured'alongthe' orthogonal passing through the point; v I

As=the total length oi said orthogonal being considk' is an expression equal to V V l? 0 c the meridianvelocity. at the point being considered, i.e., the component of the absolute Velocity in the "it will be noted all of the right hand terms are known.

c c is determined from Using these values of c and Equation 3.

1 Now using the value of c 'for any point andthe plots 1 fory, mentioned above, the-weight flow G may be calculated fr om Equation 4, all of the right'hand expressionsin which are known, e'being assumedito be,"for example 095 or'mor'e or less depending'upon an assumption'of blade thickness, number of blades'to be used, and bound ary layer thickness. This calculated value of G is now 1 checked against the assumed desired valueof G, and if "not in suf iciently close agreement, another value of c 'is assumed and the calculations 'are repeated until-"the diiierence between the assumed and'calculatedvaluesjor G'is sufiiciently small. v

Following this, values of Gfor the passages between the stream surfaces for which the meridians were foriginally assumed are calculated using Equation 4 withsuit able'limits of integration corresponding to the stream direction of themeridian of the stream surface through the point; and

' T ci =the meridian velocity at the hub end of the orthog- 'onal through the point.

The foregoing quantities are related in accordance with Equation 1 of Figure .15. It should be noted that c and cm are not separately known at this point in th calculation, by their ratio is given by Equation 1. a

In the case of an impeller, a curve of N versus 1 is :now assumed, 'Figure '11, in which N is the work input reached inthe impeller from the entrance through the distance I measured along an estimated mean meridian V such as 93 inFigure 9. (It .isto be noted that the total 'N is known from the assumed (e -r) and (c -fl see Equation 2.) I

It may also be noted that, in the case of an impeller having substantially axial inlet and substantially axial discharge, as in Figure 9, by reason of design for a variation of N Withl having asmall slope at the inlet and outlet, there will exist a substantial region in the vicinity of the inlet and inthe vicinity of the outlet Where there is no substantial change of N. Accordingly, the inlet and. outlet edges may deviate quite arbitrarily from radial planesfwithlsubstantial uniformity of N nevertheless existing along each edge. For this reason radial inlet; and

' outlet edges may be assumed as in Figure 9 for purposes of calculation, though actually these edges may, in the impeller, be non-radial. V

For the assumed inlet conditions, including an assumption that throughout the extent of the inlet the conditions he on an adiabatic line, there may now be plotted a curve of specific volume 1/ against the expression wherein N a is evaluated at the inlet, 0 is the absolutev .velocity and the variable for the plot, g is the acceleration a due to gravity, an'd'I is the mechanical equivalent'of heat.

surfaces. By trial and error assumption of jtheselimits of integration, corrected values for the intersections of the meridians with the assumed orthogonals may beldeter mined, and so there may be plotted a new set of meridians and inturn a new set of orthogonals thereto". Then all of the above calculations are repeateduntil convergency a (Note that -cos A.) s g It will be noted that c e 0, wr,.the relative velocity w and the angle 7 are all related in accordance with the vector diagram shown in, Figure 12. 'A is'known from the meridian shapes finallyffoundlj A plot of nc againstLshould give 'a flat topped curve smoothly dropping tozero at its ends correspondingto the inlet and outlet, as illustrated in,Figure 14. If such 7 shape 'of this curve is not found,'then the assumptions .as to work distributiomFigure '11, should be changed and the calculations repeated. 7 7, p The number n of blades required may, be determined from the value of 716 assuming a safe maximum value foric the local 'lift'coefiicient, e.g. 1.0 fora'n impeller and 0.5 for a diffuser.

i From Equation 5, which defines the variation of a with I along: a streamline, may be obtained byintegra tion, 45 being the wrap angle of the streamlines illustrated in Figure 7 The validity of Equation 5 will be apparentfrom consideration of Figures 8 and. 12.

- Plotting againstl or z for various streainlihsjthiee '7 01? which are illustrated inEigure l3 plots of 4; against 9 2, there is now secured the basis for fixing the actual physical construction of the blades.

Actual vanes must lie along streamlines to afiord guidance of the flow in the direction of the streamlines. The physical vanes, which, of course, have substantial thickness, may be regarded as constructed on mean surfaces which are defined by a set of streamlines belonging to the difierent stream surfaces ranging from the hub contour to the tip contour. Vanes of maximum strength would be those constructed on vane surfaces which were radial from inlet to outlet. In accordance with the above principles this end cannot be secured but rather there are provided vanes which are as nearly radial as can be attained consistent with the design principles. The departure from a purely radial vane, the mean surface of which consists of radial lines, is limited by two considera-' tions: ([1) The vane must not be subject to destructive centrifugal stresses at the normal high speeds of desired operation: and (b) the departure from radial condition is assumed to be small in the derivation of the calculation procedure above outlined. It may be pointed out that departure from radial condition may be taken into account in the calculation procedure andunder some conditions will improve the flow. Of special interest is the case of an approximation to what may be called an orthogonal flow channel for which the blade sections taken normal to the relative flow direction are orthogonal to the stream surfaces from the hub to the tip contours. Such orthogonal vanes have the property of requiring minimum lift forces to accomplish a desired work input and pressure rise and therefore lead to a maximum efiiciency. With such vanes departure from vortex flow exists, and for the attainment of this end calculation procedures may be adopted similar to those described above for the case of true vortex flow.

As will be evident, the origin for measurement of is arbitrary and consequently the streamlines as illustrated in Figure 13 may be displaced angularly in the direction of to secure a condition such as illustrated in Figure 13. in this plot the streamlines may be brought into coincidence at some such common point as is indicated at A and under these conditions it will generally be found that the streamlines on this plot Will also substantially coincide with each other'at some other point or limited region such as indicated at B. It accordingly becomes possible in many cases, as in the case of the impeller blading of Figures 2 and 3, to provide single vanes extending from inlet to outlet which have two portions either accurately or approximately lying along radial lines as, for example, at A and B of Figure 13 while at the same time departures from radial condition are minimized at other locations along these vanes.

In other cases, however, where the streamlines have considerable departures from'each other, attainment of conditions as illustrated in Figure 13 is not possible. The vanes may be split into two or more sections as illustrated in Figures 4 and 5 in which case each of the sections in itself may be caused to have one or more radial portions with minimum departure of the other portions of the section from radial condition. The several rows of blade sections thus provided are desirably secured end to end to each other to provide mutual support by welding or brazing.

The design procedure described above was-applied to orthogonals distributed throughout the lengths of the streamlines from inlet to outlet as illustrated in Figure 9. However, without material'department from the principles involved the design procedure may be applied to orthogonals restricted to'the inlet edge of a vane'as at P P and P of Figure 13 and-an orthogonal not substantially spaced from the inlet involving, for example, the points Q Q and Q illustrated in Figure 13 and along similar orthogonals at and adjacent to the outlet ends of the vanes. Under these conditions the vane shapes between these groups of orthogonals may be 10 somewhat arbitrarily assumed, nevertheless with a wor input distribution essentially corresponding to that illustrated in Figure 11 and a lift coeflicient distribution essentially corresponding to that in Figure 14. Departure from strict vortex flow, of course, is involved in this variation of design but vanes are securedwhich are unloaded at their leading and trailing portions. The 41nloading of the leading and trailing portions of the vanes is readily accomplished by starting and ending the streamlines in Figure 13 with a zero curvature in the case of suchimpellers as are illustrated in Figures 2 to 5 where the flow at the inlet and outlet has no more than a small radial component. In following the last mentioned procedures an approximate mean streamline can be found since c of the mean streamline. is approximately fixed merely by considerations of continuity. Using this mean streamline as a guide and the slopes and-curvatures resulting from calculations at the inlet and outlet there may be secured a vane shape which'is reasonable for both manufacturing and stress. As previously noted such a vane will involve departure from theoriginally assumed vortex flow.

in summary of the above it may be pointed out that there are provided in accordance with the foregoing elastic fluid passages disposed about an axis and bounded by interior and exterior surfaces of revolution, the meridian lines of at least one of which surfaces deviate substantially from parallelism with the axis, and vanes for directing the elastic fluid along flow lines havinga skew relation to the axis, the surfaces of the vanes conforming substantially to the flow lines adjacent'thereto and definable by substantial conformity at any point of a vane to the Equation 5 of Figure 15 which defines the variation of wrap angle of a vane along a streamline at any point thereof when the expressions therein are defined by Equations 1 to 4 inclusive, and the procedure outlined above. It will be noted that despite the breaking up of streamlines into possible segments, when a complete vane consists of more than one section, the value of the angle 7 is maintained 'at every point of a blade or vane element, the total of these angles thus defined being completely definitive or everyportion of the blade.

The word skew is herein used in the mathematical ense of non-confinement to a plane, particularly, herein, to a plane through the axis of rotation of the rotor. It relates to a line the definition of which requires .three dimensional coordinates expressive of position in space. Ithas the usual significance of indicating atwist about an axis more general than would be expressed by:reference to a helix of uniform lead.

It will be understood that the invention is not limited to precise conformity with theory, and intentional deviations might sometimes be desirable, particularly to minimize stresses or to secure, for example in the case of the junctionof two vane sections, a common weldingedge of as long an extent as reasonablypossibleif the angular velocity to were set at zero, it would appear that the foregoing method of design should apply to stationary passages as well as to rotating passages. However, if this were done directly in the above procedure, Equation 2 of Figure 15 would become indefinite. Consequently, this must be bypassed by assuming, instead of an N versus l distribution as in the case of an impeller, a (c -r) versus 1 distribution, o being calculated from this. The last mentioned curve is assumed of a form similar to the curve in Figure 11 but of opposite slope. Except for this change in procedure the method of design of a difiuser remains the same as for an impeller.

In view of the above it will be unnecessary to describe the correspondingsteps for diffuser .design but it may be generally pointed out that the diffuser vanes, like .the impeller vanes, are caused to conformto thecurrent streamlines which maybedisplaced and broken tip/into Pifrbm iS'I -measuredalong UV.'

o'st advantageous mechanical segments to providethe construction. I :It may be here noted that all of the foregoing matters apply'not only to'mixed flow impellers but to axial flow V impellersandj to centrifugal impellers having radial discharge, leading in both of the last cases to improvements in elficiency,

- iBy reasonfof the,fact thatjthe foregoing procedure derived' the blade shapes by a method of successive ap proximations, a final expression of. e or explicitly in 'termss'of assume-d constants was not obtained. However, by making certain approximations, there may be obtained for' the expression given in Equation 7 of Figure in which w, r, (c (c -fl and G are as heretofore defined While, referring to Figure 16, the other quantities therein may be defined as follows: a. r ST is a mean line (not a meridian of a stream surface,

though .usuallyxit .will approximate such a meridian). extending midway'between hub and tip meridians in a common axial plane. Considering anypoint P in such 15 positive, and this form is'characteristic of suchvanes.

.plane, UV is an orthogonal to the meridians of'the stream surfaces, including ,the hub and tip meridians, which passes through P. UV is then the intersection with said axial plane. of a surface of revolution orthogonal to the streamlines and-stream (surfaces.

, ;Let the'total length of.ST be l and .let l be the distance measured along ST from'S to the orthogonal UV. ,Let Anbethe total length of, UV from the inlet to the outlet ends of the vanes and let n be the spacing of i I Let-k be the curvature of the hub meridian atU and let k; 'be thecur-vature of the tip meridian at V, taken in the sensepreviously stated. a

kt+ti I 2 V V a, or. b is arbitrary but subject to the condition a+b=2,

both a and b beingpositivec vwand B are'given by the relations: laZ, l fi 2. v =the average specific volume at the inlet. V

? -v =the average specific volume at the outlet.

' A'=-the area of a surface through P and normal to the ,have an airfoil section. V

with ,whathas just been described an airfoil vane, the midtan 7, the curve, therefore, being a true-angle true-distance projection of the streamline.-= The curve S 18 typical of such plot for a streamline of an impeller for which-the 7 :prewhirl at the inlet 'is difierent from zero and the pressure ratio is of the order of 2. The curve S is typicalof such 1 plot for a stream-line of an impeller for which the prewhirl atthe-inlet'is substantiallyzero and the pressure ratio V of thevorder of 3. For any streamline, hub, tip or intermediate, such curvehas, for vanes provided in accordance 'with the-present invention, an -S-shaped form'from the inlet to the outlet inwhich thecurvature changes froma 'negative-value,through zero, to a positive value, i.'e. at

the inlet V W ,isnegative arid atthe outlet I It} should be noted that a theoretical vane of, infinitesimal .thicknessjwould follow a groupof streamlines. Onpsuch .a theoretical vanetliere is patterned the actual vane which must have-actual thickness generally varying 'frorrrhub T to tip, the actual vane therefore having as its mean surface;

the theoretical vane. Such an actual .vanewould generally This ,should not be confused line of which-would nothave the reversal of curvature abovpnoted. Such a vane, it is true, generally has, a

. reversal of curvatureof one of its actual faces, but this I is not the same the. matter of vanes built upon, or having'as a mean surface,'a su'rface made up of streamline ele-' ments having reversed curvature as described.

Here again the matters just described also apply gen 7 V erally to difiuser blades which are fastened to the casing.

stream 'surface'ineridians between the hub and tip contour component, streamlines. Through Equation 5 and a suitable displacement of the streamlines in accordance with "the description given above with reference to Figure 13 to achieve approximatelyradial conditions,; the actual :hlade surfaceslmay be derived therefrom.

1 As will appear from the foregoing, vanes may be pro- .vided in m any, instances which are very nearly radial except attheir inlet andoutlet portions but which, never- ,Since theyjare not subjected to centrifugal loads there is less restriction on this from thestandpoint of stresses and r hence substantial departures from radial condition may be tolerated;

From the foregoing it will be evident that the blade or' vane shapes for impellers and diffusers are three dimen- ,sional andof suchfq 'mthat in general machining would be prohibited, this same applying also to the blades defining turbine passages. Accordingly, a novel procedure is .adopted for forming these and for attaching them to' rotors or casings. This procedure involves die-forming the vanes from the proper metal (generally steel) from blanks of flat type which may be preliminarily machined'to provide 7 desired tapering and faired leading and trailing edges.

Thin metal may generally be used merely sufficiently thick theles s,- will} provide substantially the desired .character- .isticsofflow with'proper'blade loading with a'fair degree of approximation to vortex flow. Inaccordance with knownprinciples of aerodynamics deviations of flow may be produced byvaneswithout detrimentallyatfecting the sw rm results to be accomplished. Accordingly, impeller va'nes'riraybe rather simply defined in still another fashion pnfoi ming rather closely to the definitions heretofore set Qforthff.

f 'Figurel17 will serve to make clear-this form ofdefinition. In' this' figure, referring toFigure 8, there are plotted twotypical curves, designated S and S for streamlines :S on agraph having as ordinatesy=jfld and abscissae l. ln such 'a plot; at any point P of 'a streamline correspond- E 9 Ph -F ht? -lths-tl pe 93. ti m -wrve i to provide the necessary rigidity and strength for resist- :ance of stresses. These blanks may also be providedwith enlarged edge portions to provide the root sections of the vanes which generally have a width three or four times the average vane thickness. 7 The vanes so formed'are then sea c ured to the machined supporting surface .of the impeller or casing either by copper furnace-brazing or by flash welding, both of which are carried out in accordancewith' conventionalpracticm In the case of copper'brazing all of the vanes may be secured to a rotor or'casing simuli taneously. Copper brazing in this'fashion has been found to be extremely satisfactory, giving quite adequatestrength even in' the caseof rotor assemblies required to operate at extremely high speeds. 1

; Steel vaneshavegOOd fatigue characteristics, andsince iticular. embodiment 'ofthe invention shown by Figure 5),, ponsidering all of the stages except the fifth stage, is essen- 5 tially'a body of revolutionthe outer contour of which-has limiting amplitudes of vibration Within safe in the drawings. The total contour is made up of the individual rotors of stages 1 to 4, inclusive, which have similar characteristics and involve similar principles of design so that only one typical stage need be described. As has been indicated the rotor of each or these stages comprises a pair of radial discs connected by a tubular cylinder and providing the support for the outwardly convex shell to which the impeller blades are secured and which also provides the inner bounding surface of the diffuser of its stage. Except for the presence of the blades the impeller and diftuser portions of the shell are quite similar in structure, and, in fact, for convenience of construction may be made exactly the same so that each rotor is symmetrical about a radial plane passing through the portion of its periphery at maximum diameter. Because of the centrifugal load of the blades which it carries the impeller is usually more critically stressed than the diffuser and in the case of adoption of a symmetrical arrangement there merely resultsa diftuser portion of the rotor which is somewhat stronger than required. If equal stressing of the impeller and difi'user portions of the rotor is desired, a suitable unsymmetrical form of the rotor will result. It is, of course, advantageous to use such an arrangement of the rotor when aerodynamic and other circumstances permit.

The basic principle involved is that of providing a relatively thin flexible shell having a general form which satisfies the aerodynamic requirements and at the same time serves as a principal structural member capable of supporting its own centrifugal load plus the centrifugal load of the blades. In order to make the shell as thin and light as possible it is necessary to minimize shear and bending stresses in this part and to design it to be supported mainly by simple membrane stresses acting in the circumferential and meridional directions.

in particular to prevent the circumferential stress from exceeding safe limits in supporting a given centrifugal load it is necessary to insure that a sulficient proportion of this load will be sustained by the meridional stresses. This is accomplished by having the ends of the rotor shell anchored to the stifi frame consisting of the axial tube and the two radial discs.

In order that the shell may be of light construction, only membrane stresses should be permitted to occur therein. It will be evident that the inward resultant of the circumferential and meridional stresses is due respectively to the inward curvature of the membrane in these two orthogonal directions. The curvature in the circumferential direction is merely the reciprocal of the radius, and therefore, the greater the radius the less will be the c rcumferential curvature. Furthermore, the greater the radiusthe greater will be the centrifugal load of the shell and blades. Consequently, the portion of the total load which must be supported by the meridonal stresses is the greatest at the greatest radius and decreases with a decrease in radius. tion, therefore, varies so as to be maximum at the maximum radius, decreasing on either side. Each side of the shell gradually approaches a relatively straight, approximately conical shape, and at the innermost radii may even begin to curve outward, as illustrated, to cause the outer contour of one stage rotor to merge with that of the next one.

Since the resultant of the centrifugal load and the tangential force on an element acts exactlyin the radial direction, it is clear that the resultant of the two meridional forces must also act exactly in the radial direction. Hence, the axial components of the meridional forces on both sides of any element must be exactly equal and opposite. This means that the axial component of the meridional force is a constant all alongv the shell.

With the axial component constant, it follows that the resultant meridional force increases with the angle of inclination of the shell; hence, it is a minimum at the crown where the shell has no inclination and increases T he curvature in the meridional direc 14 rapidly toward each end. I It isusually most convenient for construction purposes to make the outerportion of the shell of uniform thickness, but because of the increase of meridional force with shell inclination a point may be reached where it becomes necessary to allow the shell thickness to increase in order to prevent excessive meridional stresses. This effect is accentuated by the factthat the decreasein radius toward the ends of the shell requires a further increase in thickness in order to maintain suihcient cross sectional area Hence, there are in general two sections of the shell: the outer part where the thickness is held constant and the meridional stress increases, and the inner part where the meridional stress is held at or below some limiting value While the thickness is allowed to increase correspondingly. In particular;

cases, one or the other of these regions may not occur.

The thickness may, for example, be allowed to increase.

continuously from the maximum radius to the minimum radius. But more commonly the meridional stress may remain within sate bounds all the way to the minimum;

radius without requiring any increase in thickness of the.

shell. The increase in thickness generally occurs in the ends ofthe rotor merely as a consequence of fairing of the contour required for aerodynamic reasons and is simply in the nature of a fillet added to the outside of the shell to join it smoothly to the overhung edge beyond the end discs. As long as it is confined to the lowerend of the shell near the junctions of the shell, the tube, the discs and overhung edge, it has no major efiect on the stress; distribution in the critical outer region. In general the parts of the rotor which extend outside the tube and discs influence the shell proper only in so far as they; actually load the discs and afifect their deflections. Hence,

alternative forms of construction may be employed tor-v these parts without greatly affecting the design of the central portion of the shell.

If the shell with its blades were not anchored at its ends to a stifleningfrarne consisting of thetube and discs, it would not sustain any meridional stresses. Consequently, the-centrifugal load would have to be supportedwholly by the circumferential stresses which would then be correspondingly high. Under this type of stress distribution the shell would undergo a relatively high radial expansion and axial shortening. The tube and discs toowould undergo some radial expansion and axial shortening, but if they were not loaded by attachment to the shell these deflections would be small compared with those of the shell. Since in the actual rotor the two ends of the shell and frameare physically connected, this difference in deflections cannot exist and hence there must be acting at these points a set of internal forces to reconcile the do flections. In particular, the axial component of the total end force is transmitted all along the shell and is everywhere constant as was pointed out above. It is this component which determines the amount of support the shell receives in the meridonal direction, and hence-also the remaining load which must be sustained by the circumferential stresses.

The tube such as 48 in Figure 3 which is under compression will generally be cylindrical although if the two ends of the rotor have difierent radii the tube may be.

conical in form or may be replaced by a number of separate struts having suitable cross sections. If a number of stages are employed, it is desirable that'the tubes of the successive hubs should be in alignment since this gives a considerable overall rigidity against lateral deflections and vibrations. The tube must sustain an axial compressive force equal to the axial component or the. tensile force on a shell where it joins the tube plus the 1 compression imparted by the tie rod such as 12 which extends through the rotor. This tube must also sustain a circumferential stress due to the centrifugal forces on it.

Since the individual blades are closely spaced around the periphery of the shell, thecentrifugalload ofsthe blades may be assumed to be uniformly distributed to 'a' mean surface of the shell.

' of the turbine rotor passages.

I the driving of generators, blowers, or the like. 7

'102 is mounted is a bearing indicated at 110, and at its 15 around the circumference of the shell. This total circumfer'ential' load mayvary. along the axis depending upon the particular dimensions of the blades at various axial positions.

The determination of the proper shell shape involves taking the foregoing considerations into account, and

performing step-by-step calculations, considering as a fundamental matter the desire to have every element of the shell; subject solely to tensile forces acting tangent These calculations are carried out in accordance with the usual principles for tail. In brief, the result may be stated to be the attainment or a rotor or hub construction comprising a member extending in the direction of and symmetrical about the axis of rotation, and in itself a body'of rotation, which is under compression, and which together with the radial enddiscs provides an anchor for a'convex shell in the form of a surfaceiof revolution which is stressed, substantially only in tension when loaded with blades. As will 'be pointed out. hereaftenthe' axial strut member Q 48 and the end discs and need not necessarily be sorather well defined from each other as: they arein Figure 3, but radial. and. axial support may be attained f throiig'h an alternative relatively rigid; central portion of the-hub of which an example appears in Figure 20. In

' this last case, as will be described hereafter, the same rotorhaving a shell construction supports both compressor and turbine blades. r s

' In the foregoing, little has been said about the applications of the principles of theinvention to turbines; It

will be evident that What has been described is directly 7 V applicable to turbine constrution inwhich the design of the turbine blades may be carried out using the. same principles-as those forming the basis for design of compressor-"blades" or vanes merely takinginto account the,

' solving stress problems; and need not be described in deingly may bemade quite thin and light notwithstanding 'the provision of adequate strength for operationat high' speeds and, on the turbine side, high temperatures. A

compressor is provided by vanes 120 carried by the hub duced for combustion through nozzleslfifl, which direct the fuel spray substantially tangentially in the direction of 'rotation'in such a manner that the largest possible flame travel is obtained, there being {provided suitable igniting means, not shown. The products of combustion fact that energy is delivered by the driving fluid to the turbinefsha'ft through the medium' of the blades rather than the reverse, so vthat the vortex strength at the inlet is greater thanthe vortex strength atthe outlet.

somewhat similar ;to those involved'in' the design of the ditfuservanes but taking into consideration the fact that the vanes 'which define the nozzle passagesmust give riseto vortexflow-suitablefor admission at the inlet ends The hub construction of a turbinerotor may also be quite similar to the hub construction for a compressor, here again so designing the hub that the shell thereof will be only under tensile stresses of the type involved in a membrane, i.e. tangent to 'tHeZmid-surface of the. shell. accordingly applicable generally to gas or steam turbines,

the considerations involved in the application of the invention to turbines will best become apparent from consideration of two examples of gas turbines illustrated in l Theno'zzleshforsucha turbine'involve design considerations While the invention is V V is discharged at the open ends of these blades to co- Figures 18, 19 and 20. The resulting gas turbine units have particular advantages which alsowill now be pointed out. t

Referring first to Figures 18 and 19, thereis illustrated 5 therein a gas turbine power plant which is particularly suitable for substantially constant speed'op'eration as for A 'shaft left hand end is spline d to the'interior of a pinion 108 of a reduction gearing system enclosed in a casing 109 and having'projectingffrom it a coupling Ill-for the de: liy'e ry of useful power. The pinion 108 is mounted in bearings 104-and 106.

. The shaftat its right-hand end is provided with a' hub comprising the radial discs 112and 114,'the axial cylindrical shell 116 and theannular shell 113 which is oflthe catenoid type heretofore described: in connection with compressors. 'This shell 118 when loaded with compressor and turbine blading, as willbe presently in- ,dicated has all of its parts under tension and accordflow through nozzle passagesdefined by blades 13} also designed in accordancewith-the principles heretofore mentioned, and flow then takes place in the form of a free vortex inthe region 134 which directs the gases into the turbine passages defined by blades 136 carried by the hub. These blades are also designed in accordance with the principles above mentioned. As will be evident,

the gases have both radially inward and axial components of flow, as-well as circumferential components offiow,

through the turbine passages and are discharged into the region 138 followed by -suitable blading 140 from which discharge is effected into the discharge passage 142. The fixed blading at 140 may be so arranged as to provide, by reducing the spin of the gases, a rise of pressure to the discharge pressure inthe passage 142 so that if this discharge occurs to the atmosphere, subatmospheric pressure may exist in the region 1 38 of dis charge fromthe turbine passages.

'As illustrated particularly in Figure 19, the blades 136 t of the turbine have a hollow form, each beingmade'of' 'two halves of sheet metal which are seam welded together along the entrance edge and along the tip contour" leaving only the discharge edge open. Passages 144 communicate with the compressor-passages so that com-' pressed air may flow therefrom into the interior of the rotor shell, this air being discharged through passages 146 into the interiorv of the turbine bladesfrom which it mingle with the discharged driving gases. As a result of this construction, effective cooling of the rotor shell 7 is providedand also cooling of the turbine blades to maintain them at a safe temperature.

Beforep'roceeding to describe the characteristics of" 7 operation and advantages of this gas turbine, reference,

will first be made to Figure'20 which shows a somewhat similar gas turbine having various features incommon with that which was just described, but adapted for variable speed operation,,this turbine being particularly useful for application to marine, locomotive and automotive propulsion.

, A hollow shaft 14s supported in shaft bearings carries" the hub 15%) of a combination compressorand turbine,

which hub is provided inthe form of a shell in the form of a solid body of revolution having the:characteristics previously, described. This hub, it will be noted differs" somewhat from the hubs previously described in that! its portions which may beconsidered the end discs merge continuously with the shell without the provision of any axial cylindrical strut spaced substantially from the axis of rotation. In this case,.the end discs merge closer to the axis of rotation and are'of a form, due to the large interior fillet construction, toprovide suflicient axial rigidity to withstand the axial stress imparted by the convex shell. The same principles, of course, apply to this hub as to the hubs previously described.

The hub portion 152 of the hub carries compressor blading at 154 constructed accordance with the principles heretofore set forth 'in detail, the complete compressor being formed by these blades and the housing receiving air from the inlet duct 155, and discharging it into the vaneless portion 156 of the diffuser and then into the vaned portion of the diffuser containing the vanes 158 also of the type previously described. The air is discharged from the diffuser into chamber 160 whence it passes through the tubes 162 into the chamber 164, and returns through the tubes 166 into the chamber 168. From this chamber the air passes through the openings 170 into the combustion chamber 172 into which fuel is introduced through the fuel nozzles 130. The products of combustion pass through the nozzles formed by nozzle vanes 174 and the passage 176 to turbine rotor passages which are defined by vanes 178 desirably of the hollow type described above in connection with Figures 18 and 19. These blades 178 are carried by the hub 150.

In the present turbine, a second stage is provided by blading 180 which may be constructed in accordance with the principles heretofore indicated, carried by a hub 182 supported by a shaft 184 which passes through the hollow shaft 148 and is supported adjacent to the hub 182 by a bearing 186 inside the hub 150, and at its other end by a suitable bearing (not shown) in the reduction gearing assembly 188 through which reduction gearing the shaft 184 drives the output shaft 190. The turbine passages defined by the blades 180 discharge the gases into the passage 192 provided with blades 194 similar to blades 140 described in connection with Figure 18. From the blades 194 the gases pass to the shell 196 whence they flow over the tubes 166 and 162 to provide heat exchange with the compressed air approaching the combustion chamber. Discharge from the heat exchanger is etfected at 198.

Cooling of the rotor 150 and of the turbine blading 178 carried thereby is effected in the same fashion as that previously described by bleeding compressed air from the compressor through the openings 200 into the interior of the rotor shell, and 'thence through the openings 202 into the interior of the hollow blades 178.

Additional cooling is effected in the designs shown in both Figures 18 and 20 by air bypassing over the crowns r of the rotors, the partitions beyond these crowns peeling off air from the compressor flows and directing the airdriving turbine and the power turbine rotate in the same.

direction, there is little or no difierential speed between the two during normal operation. Only during the starting of the plant does the condition of maximum difierence in speed between the two turbines exist since at that time when the power turbine is standing still the compressor and its turbine may be operating at full speed. In view of this fact, the hearing at 186 does not present a difiicult problem of design and may be of a conventional high speed type.

In view of the fact that the compression ratio obtainable with good efficiency by means of a single stage compressor is limited, the thermal efficiency is considerably improved by resorting'to exhaust heat recovery through the use of the heat exchanger which has been described, since then higher initial temperatures may be effectively used. However, where space saving is highly desirable, the heat exchanger may be omitted, the compressed air being delivered directly to the combustion chamber as in the modifications illustrated in Figure 18.

The gas turbine plants of the type illustrated in Figures 18 and 20 have the great advantage of maximum simplicity which is particularly apparent in the case of the type of plant illustrated in Figure 20, in which the nested arrangement ofthe turbine shafts permits 'the direct flow of driving gases from the first stage'to the second stage,--the blading of these stages being so designed in accordance with the principles given above that the vortex flow from the first stage is directly received by the second stage without any substantial loss of energy. In addition to this general advantage, there are further advantages as follows:

All surfaces of the rotor are working surfaces eliminating the win'dage and friction losses normally associated with compressors-and turbines. The elimination of these losses substantially increases turbine and compressor efiiciencies.

Excellent cooling of the turbine and its blading is obtained by direct cooling by heat transfer through the metal of the hub from the turbine to the compressor and also by the automatically forced circulation of the air introduced within the convex shell. It may be pointed out that instead of having air circulation take place from the impeller through the interior of the shell and into the turbine blading, heat transfer may also be effected by having the convex shell completely closed and filled with compressed hydrogen or another gas having similar high heat transfer characteristics.

Cooling is further elfected by the thin layer of air bypassing directly over the crown of the rotor shell from the compressor side to the turbine side through the clearance space about the periphery of the shell. This eifec tively prevents the hot gases from contacting the turbine side of the rotor hub. The internal turbine blade cooling isefiected by air bled from the compressor to the inside of the shell and thence to the inside of the turbine blades, this air discharging all along the discharge edge of each hollow blade.

By the application to the turbine design of the aerodynamicprinciples, discussed at length with particular reference to compressor blading, the highest possible turbine eficiency is obtained. In the case of the compressor, the heretofore mentioned principles aid, among other things, to obtain a substantial part of the pressure rise by means -of.the. centrifugal effect. Conversely, in the case oftthe vturbine, a substantial percentage of the energy handled by the turbine is caused by expansion against the centrifugal field, which part of the energy is converted with one-hundred percent efiiciency into mechanical energy. The turbine may, therefore, ,beaptly described as a centripetal vortex turbine.

' Another advantage of the constructions illustrated is that the stationary parts exposed to high temperature are completely surrounded by air discharging from the compressor so that the unit will be exteriorly relatively cool with the heat exchange which does occur being between the high temperature gases and the air approaching the combustion chamber so as to provide desirable heating of this air. I

What is claimed is: v r

1. An elastic fluid mechanism including a rotor comprising a hub member and vanes carried thereby, said hub member including a substantially rigid inner por-' tion, said inner portion comprising a pairof radial end discs joined and held spaced at their peripheries by rigid strut means capable of sustaining axial compression stresses, and a hollow and relatively flexible shell having as its mean surface a surface of revolution which is throughout its extent outwardly convex, said shell beingintegral with said radial end discs and having a thickness increasing from its maximum diameter to'its junctions with said radial end discs, said shell carrying the vanes and'being shaped so that its walls are substantially only under tension under the action of centrifugal forces.

2. 'A compressor for elastic fluid including a rotor havrtiallyradial with respect tothe rotor axis. 7

' planero'f said "orthogonal surface 'of revolution between ing a hub member of increasing radius from its inlet towards its outlet, and vanes carried by the'hubmemberand a housing surrounding thevanes and cooperating with the vanes and the hub'member to define elastic fluid passages, said vanes having throughout said passages their surfaces shapedtocoincide with'flow -lines of said fluid, whichfiuid'has'a flow such that at every point r from the first mentionedpassages elastic fluid having spin due to flow through the first mentioned passages, said spin reducing vanes having throughout the second: mentioned passages their surfaces shaped to coincide with 'flow lines of said fluid, which fluid has a flow such that '25 at every point of any section of said second mentioned passages orthogonal to the flow there is'a substantially constant product vof the whirl component of flow times the radius. a V

, '6; An elastic fluid mechanism comprisingmembers providing inner and outer, surfaces of revolution having a commonaxis and vanes extending between'said surfaces to define, Nvithportions of said inner-and outersurfaces,

elastiefiuid passages disposed about said axisQthe meridianlines ofv at leastone=of said surfaces deviatingifrom parallelism with said axis, and said vanes directing the elastic fluidalong flow lines having 3a skew relation to 'said axis, the surfaces of said vanes conformingto 'the flow" lines adjacent thereto .and characterized by substantial conformity at any pointof a vane to the equation' a a A -gt Marga-a I k being the curvature ofazmeridianzof:saidrinnerzsurface of revolution at its intersection with said orthogonal'sur- 7 face of revolution, and k being .the curvature of 'a:me-

ridian of:said outer surface ofrevolutionat its intersec- 0 tion with said orthogonal surface of revolution;

e is the'base of natural "logarithms;

is the average specific volume of the elastic fluid at the inlet;

v is .the average rspecific volume of the elastic :fluid at theoutlet; V a

VA is the area between-saidinner and outer surfaces of revolution-of asurface through the first mentioned point and normal to the meridians of the surfaces of revolution through "the =flow lines;

*6 is .the weightflow of .the elastic fluid;

a .andlb are positive constants related by the relation a+b ='2;:and V a and 13 areconstants, of which il $u$2 .and l'flZ;

Said vanes having their maximum aerodynamic loading between their inlet andoutlet ends. a

a 7. An elastic fluid mechanism in accordance with claim 6 in which the member providingthe inner surfaceof revolution is a .rotor rotating about said axis, and in which the vanes arecarried by the rotor.

8. .An elastic'fluid mechanism in accordance with claim 7 7 "provided with means for, driving thev rotor toimpart energy tothe'elastic fluid; a 9. :An elasticrfluid mechanismin accordance withaclaim 7 provided with 'means .for directing .the elastic fluid at high velocity to said vanes to drive the rotor.,

10. :An elastic .fluid mechanism in accordance with I providing inner and outer :surfacesof revolution having a a common axis" and vanes extending 'between said sur-:

in which: 7 i l 'y is the angle at said point which the vane makes with respect to the meridian of the surface of revolution through the flow line'at the point measured in the tangent plane tolthe last mentioned surface at the point;

a) is the angular velocity of the vanes;

r is the radius at the point;

(c -1%, is the'vortex strength at the inlet;

(c 'r) out is the vortex strength at the outlet;

m is equal to a a 7 l being thetotal 'lengthbetween-the inlet and outlet ends of the vane of a meanline extending midway between radially related meridians of said inner and outer surfaces of revolution, and 1 being the 'dista'nce'measured from the inlet end of thevane along said mean line to a surface of revolution through said point orthogonal 'to said inner and-outer surfaces-*of revolution; t t a 1 An is the total length of the intersection with an axial said inner and outersurfaces of revolution;

'11 is the length measured in 'an "axial plane through said'pointto the point from the circle of :intersection of said orthogonal surface of revolution and a surface of revolution throughsaid mean line along the intercept of faces to define, *w r 0f Said inner and outer surfaces, elastic fluid passagesfdisposed about said axis,

them'eridian lines of at least one of said surfaces deviating from parallelism with said axis, and said vanes directing the elastic fluid along'flow lines having a skew r a relation to said axis, the surfaces of said vanes conform ing to the flow lines adjacent thereto and characterized by substantial conformity at'any point of a'vane'to the equations. a

k is the curvature of the meridian of said inner stir- 21 face of revolution at the base of an orthogonal to surfaces of revolution through flow lines which orthogonal passes through the point;

.5 is the distance from the hub to the point measured along said orthogonal;

k is equal to 0 is the whirl component of the velocity of flow of the elastic fluid at the point as defined by Equation B;

G is the weight flow of the elastic fluid;

e is the constriction coeficient;

v is specific volume;

The integral of Equation C is taken along said orthogonal;

'y is the angle at said point which the vane makes with respect to the meridian of the surface of revolution through the flow line at the point measured in the tangent plane to the last mentioned surface at the point, c in Equation D being evaluated at the point;

n is the number of said vanes;

0 is the lift coetficient; and

udl is the change of vortex strength at the point with respect to distance measured to the point from the inlet end of a vane along a meridian to the surface of revolution through a streamline through the point, c in Equation B being evaluated at the point, and M having its maximum value between the inlet and outlet ends of the vanes.

12. An elastic fluid mechanism in accordance with claim 11 in which the member providing the inner surface of revolution is a rotor rotating about said axis, and in which the vanes are carried by the rotor.

13. An elastic fluid mechanism in accordance with claim 12 provided with means for driving the rotor to impart energy to the elastic fluid.

14. An elastic fluid mechanism in accordance with claim 12 provided with means for directing the elastic fluid at high velocity to said vanes to drive the rotor.

15. A diffuser for a rotary compressor, said diffuser comprising members providing inner and outer surfaces of revolution having a common axis and vanes extendabout said axis, the meridian lines of at least one of said surfaces deviating from parallelism with said axis, and said vanes directing the elastic fluid along flow lines having a skew relation to said axis, the surfaces of said vanes conforming to the flow lines adjacent thereto and characterized by substantial conformity at any point of a vane to Equations C, D and E of claim 20 and the definitions of the terms of said equations inclaim 20, said vanes being stationary, a: being zero.

16. An elastic fluid mechanism comprising members providing inner and outer surfaces of revolution having a common axis and vanes extending between said surfaces to define, with portions of said inner and outer surfaces, elastic fluid passages disposed about said axis, the meridian lines of at least one of said surfaces deviat-ing from parallelism with said axis, and said vanes directing the elastic fluid along flow lines having a skew relation to said axis, the surfaces of said vanes conforming to the flow lines adjacent thereto and characterized by the fact that, 7 being the angle at any point which the vane makes with respect to the meridian of the surface of revolution through the flow line at the point measured in the tangent plane to the last mentioned surface at the point, and 1 being the distance measured to the ing between said surfaces to define, with portions of said inner and outer surfaces, elastic fluid passages disposed point from the inlet end of a vane along said meridian,

a plot of y against 1, defined by the relation is a curve having a point of inflection between the inlet and outlet ends of the vane,

=tan 7 end. r

17. An elastic fluid mechanism in accordance with 16 in which the member providing the inner sur- References Cited in the file of this patent UNITED STATES PATENTS 2,378,372 Whittle June 12, 1945 2,446,552 Redding 1 Aug. 10, 1948 2,623,357 Binnann Dec. 30, 1952 2,785,849 Stalker Mar. 19,1957 2,801,071 Thoi'p July so, 1957 FOREIGN PATENTS 114,335 Germany Oct. 18, 1900 being negative at the inlet end and positive at theoutlet UNITED STATES PATENT OFFICE CERTIFICATE OF CORRECTION Patent No, 2,943,839 July 5, 1960 Rudolph Birmann It is hereby certified that error appears in the printed specification of the above numbered patent requiring correction and that the said Letters Patent should read as corrected below.

Column 20, line 24, for "Said" read said column 22, lines 7 and 8, for the claim reference numeral "20", each occurrence, read ll Signed and sealed this 31st day of January 1961.

(SEAL) Attest:

KARL AXLINE ROBERT c. WATSUN Attesting Oflicer Commissioner of Patents

Images