j^cfrTK'Bsi^ Kmi**a»* NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS TECHNICAL MEMORANDUM 1359 AIR -WATER ANALOGY AND THE STUDY OF HYDRAULIC MODELS By Giulio Supino Translation of *La Similitudine Aria-Acqua e lo Studio dei Modelli Idraulici." L'Energia Elettrica (Milano) , vol. XXVIII, no. 11, Nov. 1951 Washington July 1953 UNIVERf^lTY or R nR(nA ~~^ ._MT iCc LIBRARY l.U. boX 11701 i NATIONAL ADVISORY COMMTITEE FOR AERONAUTICS TECHNICAL MEMORANDUM 1359 AIR-WATER ANALOGY AND THE STUDY OF HYDRAULIC MODEIS^ By Giulio Supino SUMMARY The author first sets forth some observations about the theory of models. Then he establishes certain general criteria for the construc- tion of dynamically similar models in water and in air, through refer- ence to the perfect fluid equations and to the ones pertaining to viscous flow. It is, in addition, pointed out that there are more cases in which the analogy is possible than is commonly supposed. AIR -WATER ANALOGY AND THE STUDY OF HYDRAULIC M0DEI5 1. The use in model experiments of a fluid which is different from that of the original one has been known for some time. One is reminded in this connection that tests are made in water of bodies which are designed to operate in air (dirigibles, airplanes, railway cars, etc. ), and the first investigations in this direction were carried out by G. A. Crocco in I906 j and it is recalled that tests are made in air (that is, in wind tunnels) on configurations which are designed to operate in "La Slmilitudine Aria-Acqua e lo Studio dei Mbdelli Idraulici." L'Energia Elettrica (Milano), vol. XXVIII, no. 11, Nov. 1951. Theoretical background for experiments now going on at the Hydraulics Laboratory of the University of Bologna, sponsored by the ANIDEL (National Association of Electric Power Companies). In particular, one should consult: Crocco: Dynamics of Dirigible Balloons (Dinamica degli Aerostati Dirigibili), Bulletin of the Italian Aeronautics Society (Bollettino della Societa Aeronautica Italiana), April 1907. B. Finzi Contini: Concerning an Experimental lyfethod for the Study of Car Madels (Su di un Metodo Sperimentale per lo Studio dei Modelli di Veicoli), The Engineer (L'Ingegnere ), 1935. G. Corbellini: The engineering of rail transportation (Tecnica dei Trasporti su Rotaia). Rome, I9UI, pp. 210-232. NACA TM 1359 ■3 water, or are otherwise subjected to hydraulic loads . Within this frame of reference the Hydraulics laboratory of the University of Bologna has in progress the setting up of the necessary mechanical devices for the carrying out of certain research work, and this paper, in the form of a preview, is intended to present some facets of the possibilities inherent in experimentation that can be conducted with these two fluids. 2. Some considerations of a general nature are to be discussed first of all. The theory of mechanical similitude can be developed from two different points of view. The first of these ways of looking at the question is of an inductive sort, and it is based on the :t-theorem (of Vaschy-Riabuscinski -Buckingham), while the second viewpoint is deductive in natvire and stems from the differential equations which govern the phenomenon. Since the import of the two methods of approach each differs from the other, it is worth while to review them briefly with respect, how- ever, merely to the hydraulic problem. The expression of the rt-theorem is then as fo3J.ows: "Let us assTome that the following physical quantities are involved in the hydraulic phenomena under consideration: p density of the fluid I linear dimension which characterizes the body whose behavior is to be scrutinized U average velocity of the fluid f force (drag of the moving fluid on a given length of tube, thrust of a jet against a restraining wall or of a stream against an immersed body, drag of the fluid resisting the motion of an Immersed or floating body) 3 Tests made in air, in a pressure tunnel, to study liquid flows may be found in: G. Remenieras: Predetermination de la perte de charge dans une canalisation d'eau sous pression a partir de celle mesuree sur le meme canalisation parcourue par de I'air. Third Congress of the AIRH, Grenoble I9U9, vol. Ill, no. 2. P. Bourguignon: Predetermination de la perte de charge dans une canalisation d'eau sous pression a partir de celle mesuree sur le meme canalisation parcourue par de I'air. Controle de la validite de le methode sur la galerie d'amenee R.G.de I'usine Hydro -electri que du Pont- Escoffier. Ibid. III. 9. NACA TM 1359 g acceleration of gravity H absolute viscosity of the fluid rg nondimensional parameter or group of parameters which char- acterize the roughness of the walls r_ nondimensional parsjneter or group of parameters which char- acterize the shape of the body The connection between these quantities is to be expressed by a function of the form cp(p.i.U,f,g,n,rg,r-f.) = (1) Because this function is not to undergo any modification when the dimensional units are altered, then the unknown quantities will have to appear regrouped in such a way as to constitute pure numbers . Now all the distinct pure mombers which it is possible to construct through use of the quantities p, Z, U, f, g, \j, (of course rg and r^ are already pure numbers) are the following TJ, = pl^if N-^ f N2 = U2 Si ^3 = plU 11 (Froude ' s Number ) (Reynolds ' Number ) Thus in place of equation (l) one may write ^l(%^N2,N3,rf.,rg) = (2) This relationship is often written in a more simple manner as a, result of the fact that it is possible to rearrange it so that one of NACA TM 1359 the imknown quantities is given explicitly; for example, in the case of the force f, we have that the relationship f = pz2u2t(N2,N3,rf.,rg) (2') holds." Suppose now that it is required that some hydraulic phenomenon is to he studied by means of a model. In addition to the geometric condi- tions of similitude r = r ■^s s r^ = Tf ' (3) which must he satisfied (where primes refer to the quantities related to the model), it is also necessary to specify the further conditions of mechanical similarity given by lU U^ gl Si' = p' U' I'U' > (M and zV = P' 2 2 If the density of the fluid is p and its viscosity \i in the "basic" or original system, and if the model is to be tested in a fluid for which the density and viscosity are to be p' and \i' , respectively, the preceding conditions (k) constitute a system of relationships in the unknowns — , and — U' f ' NACA TM 1359 Let us denote by V = — the kinematic viscosity of the basic fluid, and likewise let v' denote the same quantity for the model system, then by means of equation (k) the following are obtained -L = (JL\^/3_ j^2/3 where k has been set in place of — . V In the case of the other mechanical parameters the following reduc- tion ratios apply time t f ^ .1/3 acceleration a a' = 1 volume flow Q = k^/3 pressure P P' = / P W2/3 [p'J static head h h' = k2/3 momentum /- -^ ^UU/^ work power angular velocity CO L L' = if^r^' P P' - if^r^' to - k-V3 NACA TM 1359 If air is selected as the fluid to be used in the study of a hydraulic phenomenon, the value of the ratio between the kinematic viscosities of water and air, v/v ' , cannot range out of too extensive bounds. In fact, for ordinary temperatures the value of this ratio, as a function of temperature in the two fluids , varies from a maximum of 0.13 at 0° C to a minimum of 0.05 at 30° C, as may be verified by refer- ence to figure 1, On the other hand, if one uses, as a basis of comparison, the kine- viscosity of water at at most vary from the value matic viscosity of water at 15°, the ratio between the viscosities can of water at 15° C l.lU X 10 V' of air at 0° C I.36 X 10~^ for air at a temperature of 0° C to the value V of water at 15 C I.II+ X 10" = 0.0838 of air at 30° C l.Gh X 10 -5 0.0695 for air at a temperature of 30° C. By use of equation (5), it is seen that the corresponding values of the ratio Z/Z ' are, respectively, O.I87 and O.I69, to conform with these just-mentioned values of the ratio v/ V . in the first case, that is, the scale of the model will have to be 5.35 times larger than that of the original, while in the second case it will have to be 5.90 times greater. It is clear that if the average temperature of the flow in which the original body is to operate as a water model is different from this 15°, other values for the Z/Z ' ratio will be obtained for the range of air temperatiires \inder which the experimentation is done; however, these values will not deviate greatly from the limiting values just adduced above . One may thus conclude that it will be useful to employ dimensional experiments with models in air to large scale whenever the original The NACA reviewer suggests that the exponents of 10 in the values of kinematic viscosity for water and air should be 5 and k, respectively; however, the ratios of these values are unaffected. NACA TM 1359 setup in water is of such small size that it would not permit an accu- rate examination of its behavior. On this account dimensional experiments can be employed for the study of flow from vents, from small valves, etc. where the influence of gravity and viscosity of the fluid can be taken ihto account at the same time. This mode of approach to the subject of mechanical similitude has, as mentioned earlier, an inductive aspect, that is, it is substantiated only provided that the initial hypothesis is true that the hydraulic phenomenon under examination depends only on the physical quantities enumerated in equation (l). When all these quantities (and only these ) come into play in more diverse theories, then I have already demonstrated earlier that the analogy between model and original is common to all the theories thus constructed, and thus the model test and the transfer from model to original lead to a result which, in a certain sense, does not depend upon these theories . On the other hand equations (3) and (h) are not amenable to any further polishing up; i.e., once the dependence of the function cp^ on N-|_, Ng and No, r.^ and r^ is established, no further importance can be attached to one or the other of these numbers, nor can any criterion be set up for preferring one over the other. The elimination of the quantities which may be considered to be superfluous has to come about througji the initial supposition that the phenomenon does not depend on them. If one or more quantities are cut out then also one or more of the pure numbers among the group found previously will be lacking. For exajnple, if one is concerned with a perfect fluid (for which ^ = O), then the Reynolds number woiild be missing. If the liquid is to be uninfluenced by the force of gravity, then the Froude number would be missing and if one should like to forget about the density, p, then the numbers N-]_ and No would be lacking. In this case, the number N-., defined as N2_ = — — would be inserted in place of N-]_ and No, in addition to having No still involved. 3. It now remains to ask whether it is possible to extend the range of usefulness of the air experiments by reducing the dimensions of the model. The answer is in the affirmative, because it is not actually G. Supino: On dimensional analysis and the theory of models (Su I'analisi dlmensionale e la teoria dei mode 11 i ) Two notes in the "Rendiconti della Accademia Nazionale dei Lincei", Sept. 1, 19^7- 8 NACA TM 1359 necessary to keep in the Froude number, and upon dropping the pure num- ber Ng, a much greater llherty of choice is vouchsafed. But in order to study this., it is necessary to study the hydraulic analogy from the deductive point of view, as applied to a particular one of the analogies. Let us focus our attention, for example, on the equations describing the behavior of an incompressible viscous liquid (that is, on the Navier equations ). These Navier equations are constituted of: (a) The equations of motion: In addition to the symbols already defined in section 2, the fol- lowing nomenclature is Introduced at this point. The local components of the forces acting on a unit mass are denoted as X, Y, Z which are components directed along the corresponding coordinate axes, respectively; p = pressure, |i = viscosity of the liquid, p = density, and — !i, — ^, and -^ are the "material" derivatives of the velocity dt dt dt components, that is to say, they are components of the acceleration of a little particle of liquid which, at the time t, is located at the point with coordinates x, y, and z. The symbol A stands for the operator ^ by bz Then the equation of motion is ^r bp ^ du pX i-+|iAu = p — bx. dt pY--^ + |iAv=p — by dt r, bp ,, . dw pZ ^+M-Aw=p — ^z dt (6) (b ) The equation of continuity ^ by bx. by bz ^ + ^ + ^ = (7) NACA TM 1359 (c ) The boundary conditions In reference to these, there are no uncertainties introduced here in the case of flow along walls, because it is obvious that the liquid sticks to them, and thus its velocity is equal to that of the wall, and also it is seen that the protuberances are reduced in the same proportion as required by the geometric proportionality factor. The question of "free surfaces" will, on the other hand, be considered later. It may be deduced directly from equations (6), that is to say, with- out reliance upon the rr-theorem) what the conditions are for similitude to hold in. the gravitational field. Let us consider^ therefore, the third of equations (6), applying to the original setup. Let the z-axis be directed upwards, whence Z = -g; then we have 1 ^ |i , dw £1 + _ Aw = — P ^z P dt Let p', p', z', etc. be the magnitudes related to the model conditions; then in this case we have 1 Sp' H' , dw' ■^ + — Aw' = p' ^z' p' dt' But it is true that , _ V p' = itp, p' = 6p, V k z' = Xz, w' = Xt~ w, Aw' = X, t" Aw so that upon comparison of the equation relating to the original condi- tions to that relative to the model, it follows that B\ _ 1 . — = kXT = XT-2 because g remain? unchanged in the two cases. 10 NACA TM 1359 _2 The condition that Xt = 1 is equivalent to the one denoted by 2 — = const. (Froude's no. required to remain \inchanged). The condition that kX,T = 1 when combined vlth the preceding condi- tion leads one to write ^Xt{xt-^) = kX^T-1 = 1 and this corresponds to the Reynolds number condition. Finally, the condition — =1 when combined with the condition It -2 2 Xt =1 and by making use of the relationship that « = fX leads one to write ^(Xx-^) . that is to say, we have p = const. It must be remembered that the law of geometric similtiude must be upheld, and thus equations (3) are valid. Consequently we are brought back to equations (k). It is evident that equation (7) and the condi- tions at the wall do not impose any other restrictions. k. The preceding result does not inform us of anything that could not also be derived from the Jt-theorem. But considering that this time we have within our grasp the very equations with respect to which we want to deduce the law of similitude, then it is thus possible to intimate that a deeper study of these equations leads to a reduction, in certain cases, of the number of imposed conditions. A very simple trick way of doing this is as follows, which is a well-known device, at least theo- retically (compare, for example, p. 3^ of Pistolesi's Aerodlnamica ). Let us return to examination of equation (6). Let the pressure p be divided up into two portions so that p = p-, + Vn- After doing this, we may write the equations of motion which apply in a gravity affected field as NACA TM 1359 11 ^(^1 ■" ^2) , du + n Au = p — bx dt ^(Pl ^ P2 ) ^ dv . — ^ + 11 Av = p — > ay- dt ^(^1 ^ P2) ,, , dv 7 + H Aw = p — bz dt (8) If we now substitute for p^ the value p, = -jz, then we shall wind up with nothing more than the eqiiations ^ 2 A du '' + ^ Au = p — dt bK ^2 ^ dv . + la Av = p — > by "^ dt ^ ^2 ^ dw - — + |i Aw = p — bz dt (9) and the law of similitude can be deduced by stipulating in these relation- ships that it must hold true that 5X ., 1 Xt but without the qualification that these three ratios must necessarily be equal to unity. 12 NACA TM 1359 The solution that comes most readily to mind consists therefore in selecting the Froude number in a p\irely arbitrary manner and in demanding that the following two relationships hold — ^^ = 1 and kX. t that is to say, that the N-j_ and No parameters should remain constant; so that one may write ^l(\'^3'^s'^f ) = ° If the function cp-]_ can he cast into explicit form (let us suppose it is the drag force which we want to work with) then we get f = p^V^^(N3,rf,rg) which is just a way of saying that the drag force, f, turns out to be a function solely of the Reynolds niimber and of the shape parameters rg and r^ . Nevertheless the experimental facts will have to be handled in the following way. First carry out the tests of a hydraulic phenomenon, directly based on the Navier eqxiations, by means of a model whose dimen- sions are to be obtained from those of the original configuration through It should be remarked that this trick also is of interest from a conceptiial standpoint in that it serves to sever the drag measured in tubes from the Froude number. In fact, let us write down the expression for the change in the drag in progressing from one point in a tube to another as f = 7il-^; then it follows that i = — ■|(No,r^,r„'^ = -^-^ '- — gl V 3' f' ^^ D 2g where the symbol X^No,rg^ denotes the nondimensional function which appears conventionally in all the formulae relating to drag in cylindrical tubes . MCA TM 1359 13 application of the premise that the Reynolds niomber is to remain con- stant. After doing this experimentation, the next step will be to sub- tract the value p, ' = -7'z' from the total gross pressure measured at each point of the model. It is worth pointing out that the value of p-, ' is known at every point because the shape of the model is defined. Thus one will have at hand the quantities p_ ' = p' - p ' ; and from these values the law of similarity will permit the determination of Po • The pressure existing at the corresponding point on the original body will be given by the relation: p = Pp - 7Z. In all those cases where the value of the pressure is of no conse- quence to the experiment, no corrections of any sort are called for. It may be further remarked that if the viscosity has but a minute influence on the hydraulic phenomenon under study, then the equations which govern the flow are those of a perfect fluid, and upon subjecting these equations to the same trick, one will arrive at the conclusion that the law of similarity is completely at our arbitrary disposal (provided that the pressures are computed by means of the correction factors pre- sented above, after the similarity ratios have been selected and fixed). 5. Even when one takes into account the viscosity and stipulates therefore the constancy of the Reynolds number, the artifice being exploited here renders it possible for the dimensions of the model to be reduced from those applying to the original case. Set k = — , X = — (the symbol X used here is the reciprocal of the one employed in the previous section) and let a = — . U' When this is done it, of necessity, will follow that — =1. Now k if k lies in the range O.O7 < k < O.O8 and if \ is taken as 10, then it will be possible to obtain the desired result, namely, a reduc- tion in the size of the model to l/lO of the original, by stipulating — -„ « „„ — ^„ ^v.>*«^ , ^T to ; that is, by demanding that the model ^ 1000 1000 be tested at velocities of from lUO to 125 times those experienced by the original configuration; From the ratio ^ = n = &X^t ^ it follows that p., = — 2_ P2 R^2 -2 6X It 1^ NACA TM 1359 This condition is certainly quite severe, but it is much less so than the one which might have been imagined at first thought. Suffice it to say, in this regard, that velocities in excess of 100 m/sec can be reached and surpassed in aerodynamic wind tunnels'. On the other hand, when the Reynolds number has a large value it is not necessary to insist that the two Reynolds numbers applying to the original and to the model be definitely equal. It is sufficient to require that the model Reynolds number be merely large enough to render the drag independent of this No parameter itself. In coRtrast, the creation of a water analogy to the scale of 1 to 10 obliges one to seek (in order to obtain the same Reynolds number) a velocity which is ten times greater than in the original. But the diffi- culty inherent in reaching a fluid flow velocity of 7 to 8 m/sec in a laboratory model is much worse than that of trying to produce a velocity of 100 m/sec in a wind tunnel. Besides, the experiments on models oper- ating in an air streeim can be performed with very limited apparatus (inasmuch as such apparatus will have to withstand relatively weak forces) and thus, in addition, the dimensions can be enlarged. Such advantages will make the carrying out of the experiments a yet more simple task. Finally, when one concentrates his attention solely on the inertial drag forces, the dimensional analogy may be carried out for the case of a perfect fluid field. In this case, regardless of whether one does or does not take into account the Froude number, one can take advantage of sizably larger changes in velocity by employing air-test models. 6. A single difficulty continues to bedevil the experimenter, and this blemish is not easily eliminated. This troublesome point arises in the case when any free surface or surfaces are present. The substantial difference which exists between the free surfaces present in the two cases can be explained, from a physical point of view, as follows: The jet discharge of the air stream occurs in air and the discharge of the water jet takes place likewise in air in the usual case, so that the density p is continuous across the free surface (while only the velocity is discontinuous there) for the former jet, but in the latter setup both the velocity as well as the density are discontinuous. 'Nevertheless it is important in the research we are doing to refrain from using velocities greater than about 80 to 90 m/sec in order to keep from s'orpassing by any appreciable amount the 25 percent-of-soimd-velocity borderline. With regard to this subject the reader's attention is directed to a recent paper by J. W. Ball, entitled: Model Test Using Low-Velocity Air , Proceedings of the American Society of Civil Engineers, June 1951, Vol. 77, No. 76. NACA TM 1359 15 Pursuing this line of thought further it is seen that when the water jet discharge takes place in water (as for example, for flow across a completely submerged orifice ) then there is no difference between the two cases, and the results derived as applying in air hold in addition for the case of the water flow, and vice versa (No being eq\ial in the two cases). But in general, when talking about an air jet immersed in air, the free surface constitutes an interface which separates the flow into two distinct regions A and B; in the former the fluid is in motion, while in the latter it is at rest. If we now restrict ourselves here to consideration of perfect fluids in irrotational steady flow, it follows that the actual pressure head is constant throughout the whole fluid, or P V2 z + — + — = const. 7 2g (with the flow vector being assumed directed along the z-axis) and this relationship applies validly right on the interface which separates the two regions. Now since, p, 7, and z are continuous across the surface while V-g is zero, then it follows that V. = a const. In other words, the velocity is constant on a free surface in steady flow (it being taken for granted that the free surface is interpreted as the Interface between the air to air regions), and thus it follows in addition that the pressure head is also constant. If one confines himself to examination of two-dimensional motion, then V is constant on the interface between the two regions, and in addition the stream function ^ is also constant-^. ft .... See, for example, the results of Zimm: Ueber die Stromungs Vorgange im frelen Luftstrahl . Forschimgsarbeiten VDI, Heft Z^k, 1921, and those results obtained by Citrini concerning the spreading of a liquid jet in L'Energia Elettrica , I9U6. -^The velocity V and the stream function ^ are constant in the two-dimensional case even if the motion is rotational. i6 NACA TM 1359 It is quite different in the case of a free surface between air and water. For this type of flow, the q\aantities p, g, and z are again continuous across the interface, but 7 is discontinuous. For this reason one may write that (for irrotational flow of a perfect fluid) K^A^/ W) + Z7, Z7^ = and as a result of dividing through by 7^ and remembering that V-g it follows that = 0, 2g z = z ^B ^B If one now drops out the term — inasmuch as the specific weight ^A of air is about 1/80O of that of water, then the simple result Za 2g + z = holds. Thus, in this case, now p = const on the interface; so that, whereas the pressure head is constant on the interface (or free surface ) between air and air, the pressure is constant on the interface between water and air. The two cases become identical only when the free surface is a horizontal plane-'-^. 7. In the case of steady two-dimensional flows one gets \|f = const, in both situations. On this account, if one can set up an air-to-air model for a steady irrotational two-dimensional flow wherein the interface is in geometric analogy with the corresponding interface of the original water -to-air setup, even if only along a very small section of such an interface, and if the velocities are moreover in geometric similarity, then the analogy is vouchsafed to hold throughout the flow field. In lOn ''If the velocity V^ is very large along the free surface, then the influence of the change in height z is very small; in this case also, therefore, the analogy between the air and water regimes of flow is approximately admissible. NACA TM 1359 IT fact, in this case the functions ^ and exactly correspond along the piece of the free surface. But once having the i|f -function given equal to the 0-f unction or having given the ilf -function and its nonnal derivative, thence, by means of a well-known theorem of Kirchhoff, the \|/ -function is determined throughout the whole flow field . As it has been remarked, this correspondence has to be established, however, on a free siirface or on an element of siorface immersed in the fluid - not on a fixed wall, where the condition of no-slip at the wall contravenes the irrotationality of the flow-^. 8. The conclusion from what has been set forth above is the fol- lowing: The laws of hydraulic analogy can in general be established starting out from an inductive or a deductive origin as basis for con- sideration. In the former case one makes a list (hypothetically ) of all, and only, those factors which are involved in the phenomenon. The similarity conditions are deduced then by means of the jt-theorem, and these conditions are validated through means of other well-established theorems. In the second case (the deductive one) the physical factors which are involved in the phenomenon appear in the equations which govern the theory, and with these as a basis the law of similarity is formulated (without it being necessary to bring up the jt-theorem in any shape or manner). These equations also serve to sviggest some very useful tricks. In the case of the mechanics of fluid flow these tricks have led us to establishment of the fact that experimentation performed in air on hydraulic phenomena has a quite wide field of application. Summarizing the results of sections 4, 5, 6, and 7 we can declare that it is possible to carry out experiments in air on hydraulic phenomena when one of the following conditions is conformed to: 1. The Froude and Reynolds numbers analogy is maintained (in other words the gravitational and viscous effects are taken into accoimt), when a model is made which has a larger scale than that of the original. 2. The Reynolds number analogy is maintained during a reduction in scale (relegating to an insignificant role the influence of gravity or taking it into account by means of the artifice mentioned in section U). 3. An analogy based on arbitrary ratios when the liquid caji be con- sidered to be a perfect fluid (taking into accoimt the effect of gravity by means of the artifice mentioned in section k). See Kirchhoff: Mechanik , pp. I87-I88; and Almansi: Rendiconti della Ace. Naz.dei Lincei , Sept. 1, I907, pp. 866-867. 1 ? G. Supino: Sul Moto Irrotazionale dei Liquid! Viscosi , Rendiconti della Ace. Naz.dei Lincei, Serie VIII, Vol. VI, p. 6l5, et seq. , Sept. 1, 19^9. l8 NACA TM 1359 The similarity relationship is applicable in all cases where free surfaces do not enter the problem. When these surfaces are present the cases in which similarity can be brought into play are discussed in section 6. Translated by R. H. Cramer Cornell Aeronautical Laboratory, Inc. Buffalo, New York NACA TM 1359 19 Bibliographic Notes The publications concerning dimensional analysis ;, similarity laws, and the theory of models are so niomerous that it is not possible to list them all. We wish to record here merely those of most importance and those most easily obtained. Thus we first mention the work of P. W. Bridgman, Dimensional Analysis , New Haven, 1932; also translated into German as Bridgman-Holt, Theorie der physikalischen Dimensionen , Leipsig, 1932. In addition to this basic test we wish to add the papers of: F. Marzolo, Alcune Considerazioni sui Modelli Idraulici , Giornale del Genio Civile, I917. G. DeMarchi, Omogeneita, Slmllitudine e Modelli Idraulici , Rendiconti del Seminario Mat. e Fisico di Mllano, 1933- E. Pistolesl, Omogeneita, Simllitudine, Modelli , "Fondamenti Teorici" of the Atti del II Conv. di Mat. Applicata, I9U0. A quite extensive bibliography will be found in: U. Puppini, Idraulica , Bologna, 19^7? PP- 617-618. While the less recent research work is cited and partly paraphrased in the book: Hydraulic Laboratory Practice , New York, 192 9. Moreover one should consult the various notes of the late Prof. E. Foa, contained in the issues of L'Industria , from I928 to I929, and in Politecnico for the years 1929 to 1930 (these articles advance a point of view that is a little different from that brought out in sections 2 and 3 ) • Other works, in particular the most recent one of J. W. Ball, are referenced in the notes appended to the bottoms of the pages. 20 NACA TM 1359 in Q X K '^ '>» ° ^- 2.50 o o ■D O c -^ o ^ o 2 00 H- 1.50 4J .H O "w o o CO '> o E (A O O (A '> O o E a> c o o o a: .00 0.50 ^^ VXIO* Xir - ^ ^^"-.^^ f^ (Sr >'Ar/o ^^ 5<» I0«* 15" 20*' 25* 30* Temperture , t , deg Centigrade Figure 1 NACA TM 1359 21 9 / / 8 y ^ 7 / / 6 5 4 3 / f / u / u' 2 1 20 40 60 I 80 100 Figure 2.- Froude number analogy. 22 NACA TM 1359 1500 1000 u' 500 •; 00/ / 4^ / J y / XUmox=0.lO'"/« // y rO.I25 6 y.u.. I^^Ufnox = 1.00 m/$ 50 100 Ul U'l' Fi^Tire 3.- Rejmolds number analogy — = — —. (The symbols V V differentiated by primes refer to the "air" conditions, while the unmarked ones relate to the water case. The Umax velocities are calculated to correspond to a U' value of 50 m.sec.) NACA-Langley - 7-9-53 - 1000 4Z 8 = H o e n c >-. •a o u T3 >. X ■g " H S ^ -5" I °° " " 2 - H g .3 -^ ii c = O S W iS 2 § < o z a « >" 3< C .05 BJ K m Z W > -I « HCI j3 rt en 1 = o - .2 C ■2 I XI ^ c4 n a .2 3 o ID aJ .^ o £ S "" o n - g • _ _ o o rt ■- 2 c |- n, '^ 3 • S " - >, S 3 0) 0, a; •j; o £ o £ o ^5 s J, .- c C g OJ be n 0) . o § e u a «> 2 o 3 c o q; oj C XJ g2 01 O B < < z i '-< i:^ ^ M Q,~ cm' J," eg QQ ■3 B S a < G c >. T3 O T3 >. s 0*2- c ■g " H g (J) o 3 £ 3" 0> K ac o o S o> r z c ::3 m li _- = 3sw ^ o < a - ■Sri I- 00 -- O.S; 0) eg in 3 < c . o> OT Z W > '-' - d B < < z o2. i:^ s e T3 O s< 0" CM S M 05 ■" o .ii O in t; i> c 1^ CO *- - r^ -* — ' — . ^H o c 5 = ■ ■ " ■ c O S W c« o o rf a . « K 03 Z W > -I -r M a S '^ r. a = 5 5 c 0) o ■a , HI C ■§) ^ O o a E = 2 „ 5 X 2 "> ° S ° '■" S a •n ■a c j^ rfl " S ID >. cn ■a - - « t: 41 S 2 S (U ai 3 fTt r* G C3 nl -H ^ 4, QJ Qj - to a; 5 O -H Jh oj Q oj a> "* > rt m (U •- ^ ego bfi 3 gQ S ° ■" •3 2 §• _ t: =-3 u la o c H- J— w -^ 1^ W u U. < z 8- 1 % § <^ >." CM ^H S2. ^2i H B CO s s •5 5 S c ^ •a 8 T3 S ■< 33 0) 5^ a* 2- "5 -^ o lb i; H I .2 -" ii - £ c O S W d rt o <: 'Q . » ?. Q. V; 0) CM in cs ac tn Z W > -^ •^ « H B ■a , HI Ci s a J. £ bo o 5 = 5 * > 0) ^ 3 OJ cj rt i; .= 3 la 2 T3 o a u H TD :»■ QJ =2 2 to ^ ■. * ii 3 r! t» !" 5 m -S -S c ■a — = t: u £ 3 = P:S||2f o 5L4 w k4 w P T!l I- 3 « 13 3 0) H £ bO '00 P c T3 -S'S ^ rt " o (U o c Ot Qi t4 < < §■ c 3 3 N ■? M J," e>) J," eg y, ' — T -I e = c c >. •a o ■o >. X c H g rt rt 10 ^ < U h3 3 c CO 71 >> CO -j a, o ,„ o c S o* = -p ° J3 en (U (U 5 en e j= 0) L. H 01 0) U) •n CO t1 a S o I :^ £ ^ ? - O £ 5 rt •2 - § : 8 *2 » m '- 3 O £ 2 e d ^ * ^ 3 "I ,0 IK 2 2-a S o n o £ •" " S o ^ "35 i| C g 0) t. be en Qj Q« .ii ^ ■S 2 * a .s §- CO " " i £ I g u (1) m u "^ rt c cit V 5 cu ^ O T3 >. X - o " - 2 S > - o en ■- Z H g H 1 .3 - i; ■a rt o p ^ ■Ql "■ o nl 2 S O QJ »H c t. oT ^ cu^a>^y'--■ «hH ■< o < z ^: £ = (U a < c >^ Sr-, 0) « O" O) H c CO g H s c >> f ) T3 rt U 0) T3 >. (i> a a < oTeM ■2*2. «- > S « OS -H :g (u G ^ •" *^ _r H S .2 " i! - c< o o ^ 'a . I'' 4? .2 Cl t- 00 -H « ^ 3 < c .03 K K 03 z H > ^ ^tiS U 5 q; -^ Z Z < Q . >> m ; — 0) ' cj w o Si CO c 0) >, 01 3 5 •0 V CO ■s ■^ ■o r ? c 3 ■3 ° m .D _ 1 2 u s CO til ■a c c •a c 3 O" .2 g s E CJ 1 2 * 3 u c ■^ t CO -1 0) Z 4) n a CO m OJ ^ ,^ c _ be 3 CO I, CJ c o c o -^ l||§ll| Qj C H QJ t. ^ c3 J- a> .::! u oj o c