You are on page 1of 195

Bifurcations in Flow Patterns

Some applications of the qualitative theory of differential equations influiddynamics

Proefschrift
ter verkrijging van de graad van doctor aan de Technische Universiteit Delft, op gezag van de Rector Magnificus, prof. dr. J.M. Dirken, in het openbaar te verdedigen ten overstaan van een commissie door het College van Dekanen daartoe aangewezen, op dinsdag 31 mei 1988 te 14.00 uur.

door

Pieter Gerrit Bakker vliegtuigbouwkundig ingenieur geboren te Groningen

TR diss 1632

Dit proefschrift is goedgekeurd door de promotor Prof. dr. ir. J.W. Reyn

Aan Alida

Abstract
In the present thesis the qualitative theory of differential equations' is used along with topological considerations to discuss problems in fluid dynamics and gasdynamics. Special attention is given to the qualitative aspects of flow fields, in particular to the geometry, the shape and the structural stability of streamline patterns. The theory relies much on critical point analyses and on bifurcations in vector fields. Local solutions of the flow equations are derived to discuss changes in flow topology in conjunction with bifurcations of critical points. The theory is applied to topics of inviscid, nonlinear conical flows and of steady viscous flows over plane walls.

Contents
Page

Preface Chapter I Some elements of the qualitative theory of differential equations

1. Phase space representation of a dynamical system 2. Phase portraits near singular points 3. Topological structure of phase portraits, structural stability, bifurcation 4. Higher-order singularities in R2 5. Bifurcation of vector fields, unfoldings 6. Center manifolds 7. An approach to physical unfoldings in flow patterns 8. References

5 9 13 20 27 33 4l 43

Chapter II

Topology of conical flow patterns


45 45 46 48 51 55 55 56 61 61 66 70 75 75 76 79 83 90 92 92 95 98 104 109 112

1. Introduction 1.1. Concepts and definitions 1.2. A survey of conical flow theory 1.3. Conical streamlines, conical stagnation points 1.4. Transition phenomena in conical flow patterns 2. Local conical stagnation point solutions in irrotational flow 2.1. Conical potential equation 2.2. Conical stagnation point solutions 3. Classification of conical singular points in conical flows 3.1. First-order conical stagnation points 3.2. Irrotational attachments and separations 3.3- Higher-order conical stagnation points 4. Analytical unfoldings in conical flows 4.1. Bifurcation parameters 4.2. Approximate solutions near regular points 4.3. Saddle-node bifurcation 4.4. Bifurcation of topological saddle point 4.5- Bifurcation of topological node 5. External corner flow; a nonanalytic unfolding of a starlike node 5.1. The flow around an external corner 5.2. Boundary conditions and bifurcation modes 5.3- Bifurcations of the starlike node 5.4. Symmetrical external corners 5.5- Transition of oblique saddle to starlike node 6. References

-iii-

Chapter III Topological aspects of steady viscous flows near plane walls
1. Local solutions of the Navier-Stokes equations 2. Steady viscous flow near a plane wall, elementary singular points 2.1. Approximate solutions near a plane wall 2.2. Elementary singular points located at the wall 2.3- Elementary singular points in the flow 3. Higher-order singularities in the flow pattern 3.1. Higher-order singular points in the flow field 3.2. Higher-order singular points on the wall k. Unfolding of a topological saddle point of the third order 4.1. Local phase portraits of the unfolding k.2. Incipient bubble separation 4.3. Separation along a moving wall k.k. MRS-criterion for separation in flows along a moving wall 4.5- Unfolding model for moving wall separations 5. Unfolding of a topological saddle point of the fifth order 5.1. Description of the unfolding 5.2. Bubble capturing by a secondary separation 6. Unfolding of a saddle point with three hyperbolic sectors in a half plane 6.1. Universal physical unfolding 6.2. Bifurcation sets, flow patterns 7. Unfolding of a saddle point with two or four hyperbolic sectors in a half plane 7.1. Universal physical unfolding 7.2. Determination of codimension 7-3- Neighbouring singular points, local bifurcation sets B and B 7.4. Flow patterns and global bifurcation sets B . and B _ 8. Viscous flow near a circular cylinder at low Reynolds numbers 8.1. Description of flow topology 8.2. Symmetrical bifurcations 8.3. Asymmetrical bifurcations, transition scenario's 9 References

Samenvatting About the author

-1-

Preface
The main idea of the present thesis is to demonstrate that the qualitative theory of differential equations, when applied to problems in fluid- and gasdynamics, will contribute to the understanding of qualitative aspects of fluid flows, in particular those concerned with geometrical properties of flow fields such as shape and stability of its streamline patterns. It is obvious that insight into the qualitative structure of flow fields is of great importance and appears as an ultimate aim of flow research. For, qualitative insight fashions our knowledge; it serves as a good guide for further quantitative investigations. Moreover, qualitative information can become very useful, especially when it is applied in close correspondence with numerical methods, in order to interpret and value numerical results. A qualitative analysis may be crucial for the investigation of the flow in the neighbourhood of singularities where a numerical method is not reliable anymore due to discretisation errors being unacceptable. Up till now, familiar research methods - frequently based on rigorous analyses, careful numerical procedures and sophisticated experimental techniques - have increased considerably our qualitative knowledge of flows, albeit that the information is often obtained indirectly by a process of a careful but cumbersome examination of quantitative data. In the past decade, new methods are under development that yield the qualitative information more directly. These methods, make use of the knowledge available in the qualitative theory of differential equations and in the theory of bifurcations. The qualitative theory of differential equations as applied to dynamical systems of two-and three-dimensional vector fields appears to be very useful, in order to determine the topology and structural stability of streamline patterns, occurring in fluid dynamics. The theory originates from the work of Poincar (1880) and is further developed by Birkhoff ( ~ 1927). Lyapunov ( - 19*9). Andronov (1937) and his co-workers, Arnold (1963) and many others. For references see Chapter I (Guckenheimer and Holmes (1983))In the present thesis this theory is applied to flows, which can be described by two variables x and y and the streamlines are the solution curves of a system of the second-order, -^ = P(x,y), -^ = Q(x,y), where P and Q are related to the components of the velocity.

-2-

A survey of some elements of the qualitative theory of differential equations i s given in Chapter I . The review introduces and discusses in some d e t a i l the most important concepts, theorems, and methods t h a t are used in the subsequent analyses; for rigorous mathematical proofs the reader w i l l be r e f e r r e d to the l i t e r a t u r e . Actually Chapter I may be consulted in order to be informed about notions as phase p o r t r a i t s of a dynamical system, singular p o i n t s , t o p o l o g i c a l s t r u c t u r e , s t r u c t u r a l s t a b i l i t y , degenerate systems, bifurcation of vector fields, unfoldings, co-dimension and center-manifold theory. The f i r s t a p p l i c a t i o n of the q u a l i t a t i v e theory in this thesis, concerns the flow geometry of three-dimensional inviscid conical gas flows. I t i s t r e a t e d in Chapter I I . Conical flows have the specific property that the velocity of the gas p a r t i c l e s and the quantities, defining the s t a t e of the gas, e . g . pressure density and temperature, are constant along rays originating from a common point (conical center) . A three-dimensional conical flow i s e s s e n t i a l l y two-dimens i o n a l and i s decribed adequately on a unit sphere (around the conical center) on which the flow geometry i s displayed by conical streamlines. Conical streaml i n e s are- defined as follows. For conical flows, the family of s p a t i a l streamlines passing the same ray form a conical stream surface with v e r t e x in the conical c e n t e r . The i n t e r s e c t i o n of a conical stream surface with the unit sphere i s a conical s t r e a m l i n e . Along a conical streamline the entropy i s constant or i t jumps i f a shock i s passed. The conical flow geometry on the u n i t sphere i s governed by a second-order dynamical system. The corresponding vector field i s determined by the cross-flow velocity, being the velocity component tangential to the unit sphere. In points on the u n i t sphere where t h i s component vanishes the flow is purely radial; these points are called conical stagnation points and appear as singularities of the second-order dynamical system governing the conical streamlines. In Chapter I I , possible flow patterns near conical stagnation points are studied and t h e i r structural s t a b i l i t y i s examined. Transitions to a different conical flow pattern are interpreted as b i f u r c a t i o n s of s t r u c t u r a l l y unstable higher-order conical stagnation points. Chapter I I gives a complete c l a s s i f i c a t i o n of f i r s t - o r d e r conical s t a g n a t i o n p o i n t s and s t a r t s with the c l a s s i f i c a t i o n of higher-order conical stagnation points, by considering second- and t h i r d - o r d e r p o i n t s , these p o i n t s being of most practical i n t e r e s t in real flow problems. Chapter I I ends up with a qualitative examination of the i n v i s c i d flow around three d i f f e r e n t conical bodies: circular cones, delta-wings with arrow-shaped cross-section and external corners.

-3In Chapter I I I the qualitative theory will be applied to steady two-dimensional incompressible viscous flows along the surface of a plane or along a s l i g h t l y curved w a l l . A second-order dynamical system, whose t r a j e c t o r i e s represent streamlines, i s derived. A s i n g u l a r - p o i n t - a n a l y s i s i s performed in order to obtain detailed information about the local flow topology. In p a r t i c u l a r , s i n g u l a r points on the wall surface are of i n t e r e s t , because in these points the shear s t r e s s vanishes i n d i c a t i n g flow s e p a r a t i o n or flow attachment. As in Chapter I I the singular points are d i s t i n g u i s h e d i n t o f i r s t - o r d e r and higher-order s i n g u l a r i t i e s , the former appear only as centerpoints and saddle points. First-order saddle points that are located at the wall surface represent s o l u t i o n s that are recognized as the classical Oswatitsch-Legendre solution for flow separation or attachment. The Oswatitsch-Legendre solution i s s t r u c t u r a l l y s t a b l e i n the sense t h a t a n a l y t i c a l p e r t u r b a t i o n s in streamwise pressure gradients and/or shear stress gradients will not affect the flow topology near the separation (attachment) point. Apart from the structurally stable solutions, s t r u c t u r a l l y u n s t a b l e s o l u t i o n s appear as w e l l , they occur as higher-order or degenerate singularities of the dynamical system that governs the streamline pattern. The study of these degenerate s i n g u l a r i t i e s ; their unfoldings and bifurcational behaviour, will be the main subject of Chapter I I I . Unfoldings, r e s u l t i n g in d e g e n e r a t e s i n g u l a r i t i e s f a l l i n g a p a r t i n t o a number of f i r s t - o r d e r s i n g u l a r i t i e s , describe p o s s i b l e s t r u c t u r a l l y s t a b l e flow p a t t e r n s in viscous incompressible flow. These patterns, being more complex for singularities with a higher degree of degeneracy, are interpreted physically. Some of them appear to be very common in aerodynamics, others are new and are concerned with topics in laminar flows as: - genesis of laminar separation bubbles - flow separation on moving walls (Moore-Roth-Sears criterion) - interference of separations and attachments, and - formation of asymmetric standing eddies in the near wake behind a body (vortex shedding).

-5-

Chapter I Some elements of the qualitative theory of differential equations

1. Phase space representation of a dynamical system


With the aim of easy reference we will give in this chapter some elements of the qualitative theory of dynamical systems which we will use in the next chapters. The theory of dynamical systems has been extensively studied over a long period of time. Although many questions remain as yet unsolved, a large amount of results has been obtained and is available for applications. Numerous textbooks, articles and papers testify of the progress made in this branch of mathematics. A comprehensive survey of important achievements, including recent developments and advanced methods, is given by Guckenheimer and Holmes (1983): 'Nonlinear Oscillations, Dynamical Systems and Bifurcations of Vector Fields'. This book, written with a strong view to combine pure mathematical reasoning and applicability to practice, has inspired me in writing this chapter. The theory of dynamical systems aims to study the time behaviour of evolutionary systems which are described mathematically by an equation of the form:

where X = X(t) e R

is a n-vector and f(X,t) is a sufficiently smooth function x R.

defined on some subset U R

In applications t is usually interpreted as time. The function f defines a vector field in the n-dimensional space the so-called phase space. The solutions of the system, representing sequential states of the evolutionary process, appear as integral curves of the vector field in the phase space. In the literature on this subject, one encounters several names for these curves, we quote: solution curves, trajectories, (phase-)paths and orbits. In the same way, the terms phase portrait, phase pattern, flow or trajectory pattern are used to indicate the whole set of solution curves in the phase space. Dynamical systems where f does not explicitly depend on time are called autonomous and the trajectories of such systems do not change if time goes on. In the following we only need to study these autonomous systems, being expressed by the equation

x = f(x),

x e Rn

(1.1)

In a c l a s s i c a l treatment of such a system, see for example Coddlngton and Levinson (1955). the attention i s mainly d i r e c t e d to the p r o p e r t i e s of i n d i v i d u a l s o l u t i o n curves and culminates into questions about explicit time behaviour and dependency on i n i t i a l conditions of the solutions. A d i f f e r e n t approach t o the study of dynamical systems i s obtained i f one considers families of solution curves; then qualitative questions arise such as: do t h e r e e x i s t steady and/or p e r i o d i c s o l u t i o n s , what are t h e i r s t a b i l i t y properties. Domains in the phase space where f (X) i s a nonvanishing vector function are called regular; the phase p o r t r a i t s in a regular domain are relatively simple as they can be mapped by a homeomorphism onto a family of parallel trajectories, see Fig. 1.1.

Fig. 1.1. Mapping of the paths in a regular domain into a field of parallel paths. An important c l a s s of s o l u t i o n s of Eq. (1.1) are the so-called equilibrium or steady solutions: X = X with X satisfying f(X ) = 0. An equilibrium or steady s t a t e s o l u t i o n i s thus represented in the phase space by a single point and corresponds to a c r i t i c a l point of the vector f i e l d , in the following such a point will often be referred to as a singular point of the differential equation (1.1). Steady solutions: X= X cannot be reached by a neighbouring s o l u t i o n X(t) in f i n i t e time. A steady solution i s stable if a solution X(t) through a point in a neighbourhood of X , remains close to that steady solution for t + . The steady s o l u t i o n i s asymptotically s t a b l e i f a l l neighbouring solutions

-7X(t) X as t < . Steady solutions not satisfying the stability condition are

said to be unstable. Some examples of stable and unstable solutions in the phase space are depicted in Fig. 1.2.

a:stable

b:asympfoficatty sfable

^unstable

Fig. 1.2. Stable and unstable steady solutions. The phase portrait of trajectories near singular points can be rather complicated and is usually not homeomorphic with a field of parallel trajectories. Important elements of phase portraits of systems are singular points and the local trajectory pattern near these points. Therefore, a systematic description of phase portraits near singular points is a useful tool when analysing phase portraits of systems. A detailed treatment of phase patterns near singular points is given in paragraph 2 of this chapter. Another class of solutions of Eq. (1.1) we want to mention here are the socalled periodic solutions satisfying: X(t+T) = X(t) with period T. Periodic solutions of a dynamical system appear as closed paths in the phase space. Closed paths in R2, representing periodic solutions of two-dimensional systems, must divide the phase plane into an inner and an outer region. If there exists a neighbourhood of a closed path which does not contain another closed path, this path is called a limit cycle. A limit cycle is stable if all neighbouring paths of the limit cycle approach it if t >< > , otherwise the limit cycle is unstable: see Fig. 1.3. An important question, but often difficult to answer, is to determine whether an autonomous system has periodic solutions and where they appear in the phase space.

-8-

a:stable limit cycle

b:unsfable limit cycle

F i g . 1.3. Limit cycle i n R1 An e l e m e n t a r y c o n d i t i o n t h a t can be used for the non-existence of closed paths i s formulated by Bendixson's c r i t e r i o n . Bendixson's c r i t e r i o n Let x = P ( x , y ) and y = Q ( x , y ) be an a n a l y t i c a l p l a n e on which the divergence of t h e v e c t o r f i e l d : a r e no closed paths l y i n g e n t i r e l y i n U. The proof of Bendixson's theorem goes as follows. A p p l i c a t i o n of the divergence theorem along a closed curve T l y i n g e n t i r e l y i n U gives dynamical div(P.Q) s y s t e m and l e t U be a s i m p l y - c o n n e c t e d domain of the phase d o e s n o t change s i g n and i s not i d e n t i c a l l y z e r o . Then t h e r e

;/
UY

i * ^
3X 8y

dx dy

= ' ( P - Q > n ds
T

with U

the interior of T, n the outward normal and ds a line element of T.

If 7 is a path of the vector field (P,Q), then (P.Q) and n are perpendicular so that the line integral I vanishes identically. But since the integrand of the T integral II is of one sign, the integral cannot be zero. Then the curve T cannot U T be a path of the vector field (P.Q).

-92. Phase portraits near singular points n Assume t h a t a dynamical system X = f (X), X e R has a singular point X so that f(X ) = 0 . In order to characterize the t r a j e c t o r y p a t t e r n near X we assume f(X) to be s u f f i c i e n t l y smooth and we expand f(X) near X . Retaining only the linear part in the expansion there follows with X - X = E o 5 = Df(XQ) 6. 5 e Rn (1.2)

3f . where Df(X ) denotes the Jacobian matrix [-] o ox . J derivatives of the vector function ^ ( x j . . . . - xn) f 2 ( x r . . . xn)

of the f i r s t o r d e r p a r t i a l

n(xl'

s/

evaluated in the singular point X Equation (1.2) is a linear system with constant coefficients and can be analysed with classical methods yielding the corresponding 'linearized' phase portraits. However these, 'linearized' phase portraits are not necessarily equivalent with those near X of the original non-linear system. The relation between both phase portraits is given by the theorem of HartmanGrobman which holds for systems in R . Hartman-Grobman If Df (X ) has no eigenvalues with zero real part, then the family of trajectories near a singular point X of a nonlinear system X = o f(X)* and those of the locally linearized system have the same topological structure; which means that in a neighbourhood of X there exists a homeomorphic mapping which maps trajectories of the non-linear system into trajectories of the linear system. When Df (X ) has no eigenvalues with zero real part, the singular point X is called hyperbolic or nondegenerate.

* In this thesis we restrict ourselves to the application of cases where f is analytic but the theorem can be extended to a wider,class of functions.

-10The theorem of Hartman-Grobman implies that local linearization near singular points will be an effective method when analysing phase p o r t r a i t s of n o n - l i n e a r systems. Therefore i t i s worthwhile to recall f i r s t the theory of linear systems in some d e t a i l . Linear systems Linear systems in R are t r e a t e d e x t e n s i v e l y and in various aspects in the literature. Especially, systems in R2 are very well known and receive much attention in many textbooks on ordinary d i f f e r e n t i a l e q u a t i o n s , e . g . Coddington and Levinson (1955) and Jordan and Smith (1977). Furthermore, for l i n e a r systems in R' reference may be made to Reyn (1964), where a sound treatment of s i n g u l a r points culminates i n t o a conveniently arranged survey of three-dimensional phase p o r t r a i t s . Let us briefly recall here the main results for l i n e a r systems in R2 , as they are frequently used in the following chapters. Consider the two-dimensional system

where A i s a constant 2x2 matrix. The system has only one singular point: the origin (0,0). A c l a s s i f i c a t i o n of phase p o r t r a i t s can be g i v e n i n t e r m s of t h e t r a c e (p = A.+A_) and Jacobian (q = A..A_) of A; A., A_ being the eigenvalues of A. W e consider several cases (i) q < 0: the eigenvalues A. and A_ are real with different signs. The singular point i s a saddle point and i s unstable. 0 < q < (j-p)2 : the eigenvalues A. and A _ are real, unequal and of the same sign. The singular point i s a stable node (sink) i f p < 0 and an u n s t a b l e node (source) if p > 0.

(ii)

( i i i ) q > (j-p)2 (p * 0 ) : the eigenvalues A- and A_ are conjugate complex with non zero r e a l p a r t . The phase portrait shows a focus or spiral near the singular p o i n t . The s i n g u l a r point i s a s t a b l e focus i f p < 0 and an unstable focus if p > 0.

-11(iv) q > O, p = O: the eigenvalues A called a center, (v) q = (S-p)' (p * 0 ) : the eigenvalues A. and A_ are real and equal (A.. = A_ = A ) . The nature of the phase portrait depends on the Jordan form of A. If A has the Jordan form ( _ ) the trajectories form a star shaped node, which .) the phase portrait is called an inflected is stable for p < 0 and unstable if p > 0. If A has the Jordan form L parallel at infinity. (vi) q = 0, p * 0: one of the eigenvalues is zero; the phase portrait consists of a family of parallel paths and a line of singular points, which is stable for p < 0 and unstable if p > 0. This classification of linear systems is conveniently arranged in the p-q-plane, shown in Fig. 1.4. node; all trajectories tend to the origin in the same direction and are and A ? are purely imaginary. The trajec-

tories form closed curves surrounding the origin and the singular point is

q = XvX

p = X, + \2

Fig. 1.4. General classification of a linear system in R*.

-12The isolated singular points for a linear system are saddles, nodes, foci and centers. Since for nodes, f o c i and s a d d l e p o i n t s Re(A) * 0 h o l d s , Hartman-Grobman's theorem implies that adding sufficiently smooth non-linearities does not change the phase p o r t r a i t s near these types of singular points. Centers i n n o n - l i n e a r systems satisfy Re(A) = 0, and t h e i r existence cannot be shown by l i n e a r i z a t i o n . As an i l l u s t r a t i o n consider the nonlinear system
x = y
x2 y

y = -x +

with eigenvalues A , A_ = i. Unless e = 0, the singular point (0,0) is not a center as in the linearized system, but a nonlinear focus, stable if E < 0. Nonhyperbolic singularities as characterized by a vanishing Jacobian (q = 0) can appear in the phase space in different forms: isolated points, curves etc. The isolated points are usually denoted as multiple-equilibrium points, degenerate singularities or higher-order singularities. The topological structure of the local phase portrait near higher-order singularities can be very complicated and a general classification for singularities in R is hard to give. For R2 , Andronov et. al (1973) offer a classification for isolated higher-order singularities having a nonvanishing degenerate linear part. These points will be reviewed in more detail in paragraph k.

-133.Topological structure of phase portraits, structural stability, bifurcation 2E22S2? trucyE_2?_E!}&_P2E&iE The concept of topological s t r u c t u r e of phase p o r t r a i t s has a l r e a d y been introduced in paragraph 1 when we mentioned the fundamental theorem of HartmanGrobman. Now a more thorough examination of this concept will be given because i t enables us to introduce and explain the concept of structural s t a b i l i t y . C h a r a c t e r i s t i c f e a t u r e s of the phase p o r t r a i t which may be called qualitative properties are for example the number and type of singular points, the existence of closed paths and regions of attraction. Formally, one may define qualitative properties as those properties of the phase p o r t r a i t which remain i n v a r i a n t under a topological mapping or homeomorphism. A topological mapping between two regions in de plane i s a one-to-one and bicontinuous mapping, meaning t h a t each point M i s mapped exactly onto one point M' and that d i s t i n c t points M. and M? are mapped onto d i s t i n c t points M' and Mi, and the mapping i s continuous e i t h e r way. An intuitive description of a topological mapping of the plane onto i t s e l f i s given by Andronov e t . al (1973) as follows: 'imagine the plane i s to be made from rubber which i s deformed in some way, stretching and squeezing i t at various points, but without tearing or folding. Any topological mapping of the plane into i t s e l f i s either a deformation of the above type (without tearing and folding) or a mirrorreflection of the plane followed by such a deformation'. I t may be c l e a r t h a t a topological mapping of a phase p o r t r a i t can result in d r a s t i c changes in the shape of t r a j e c t o r i e s , but c e r t a i n p r o p e r t i e s w i l l n e v e r t h e l e s s be preserved. As an example: s i n g u l a r p o i n t s are mapped into singular points and closed curves remain closed curves. The concept of topological structure of a phase p o r t r a i t in R2 i s now indirectly indicated by the following definition. Consider in a region G d f f two dynamical systems D. and D_, given by X = f^X) x = f 2 (x) (D1)
<D 2 >

-14The phase portraits of the systems (D.) and (D_) have the same topological structure if there exists a topological mapping (homeomorphism): T which maps G onto G and which takes paths of (D.) over into paths of (D_)*). If two points X. and X ? lie on the same path of system D., then their images TX. and TX_ lie on the same path of system (D_). Also, if two points X1 and X_ lie on the same path of system (D_), then their - 1 - 1 images T X. and T X_ lie on the same path of system (D.). The given definition of topological structure is in a certain sense indirect since it does not state exactly what topological structure is, but it specifies the necessary conditions for equal topological structures. Structural_stability; bifurcation After the definition of equivalent topological structure, we now introduce the concept of structural stability. Consider the dynamical system (S) defined in a region G C R . X = f(X) (S)

System (S) is said to be structurally stable if an infinitesimal change of f(X) leaves the topological structure unaffected in G, otherwise the system is called structurally unstable. It should be noticed, that structural stability is not an intrinsic property of a topological structure but is related to the class of infinitesimal changes of f (X) that are allowed. Both the class of perturbations of the vector field which are admitted and f(X) itself determine whether there is structural stability or not.

') If the topological mapping T (with inverse T

) is k times differentiable

(k > 0), the mapping is called C diffeomorph and the two vector fields are k k said to be C -equivalent; C -equivalence with k > 0 implies that certain smoothness properties (k-times differentiable) of trajectories remain preserved by the mapping process.

-15This c h a r a c t e r i s t i c property of s t r u c t u r a l s t a b i l i t y i s i l l u s t r a t e d i n the next example where two d i f f e r e n t system (in R* ):


x = -x

types of perturbations are imposed on the linear

(1.3) y = -2y which has a node a t the origin. F i r s t we c o n s i d e r an a n a l y t i c p e r t u r b a t i o n by e.^ing a linear term ux to the right-hand side of y:
x = -x

(1.4) y = -2y + ux with u being a small perturbation parameter. The origin is a stable node for u = 0 as well for u * 0, thus (1.3) is structurally stable with respect to the perturbations of the (analytical) character given in Eq. (1.4). Next we consider a non-analytic perturbation*) of (1.3) s y adding a perturbation of the form u /|x| x = u /|x| - x (1.5) y = -2y If l i > 0 system (1.5) has two singular points on the x-axis: (0,0) and (u2,0). Near (0,0) the trajectories behave partially like those near a stable node and partially like those near a saddle point. At (uJ,0) there occurs a stable node. For i i < 0 similar results will follow. The phase portraits obtained if u varies near zero are shown in Fig. 1,5.

*) Although such non-analytical perturbations seem a bit far-fetched, they appear in actual flow situations, as can be observed in Chapter II where structural stability in conical flows around an external corner is treated.

-16-

=0

|j>0

F i g . 1.5. Nonanalytic p e r t u r b a t i o n of a nodal p o i n t . The examples i l l u s t r a t e perturbations t h a t t h e s t a b l e node a t (0,0) i s s t r u c t u r a l l y s t a b l e against the a s g i v e n by Eq. ( 1 . 5 ) . E v i d e n t l y , structural

a g a i n s t a n a l y t i c a l p e r t u r b a t i o n s (Eq. ( 1 . 4 ) ) b u t s t r u c t u r a l l y u n s t a b l e (nonanalytical)

s t a b i l i t y of a t o p o l o g i c a l s t r u c t u r e cannot be e s t a b l i s h e d i n d e p e n d e n t of c l a s s of p e r t u r b a t i o n s t h a t a r e imposed on the dynamical system.

To g a t h e r more a s p e c t s about s t r u c t u r a l s t a b i l i t y and l e a v i n g the d i s c u s s i o n s as simple as p o s s i b l e , l e t us c o n t i n u e by c o n s i d e r i n g systems i n R2 i n some more d e t a i l . Assume t h a t the l i n e a r system:
A X

X 6 R2

(1.6)

is perturbed by adding a term u f(X) yielding the nonlinear system X = A X + u f(X) u R

(1.7)

where f(X) at least Cl and f(0) not necessarily equal to zero. If the eigenvalues A-, A_ of A have non-vanishing real parts, Re(A. _) * 0 then it can be shown that near the origin the phase portrait of (1.6) is structurally stable against the C1-perturbations given in Eq. (1.7). To this end, the phase portrait of Eq. (1.7) near the origin has to be determined. A singular point: X of Eq. (1.7) satisfies AX +K u f(X ) = 0 o' = -u A -1 f(X ) the i m p l i c i t function for

Since A is an invertible matrix, and X t h e o r e m can be u s e d t o f i n d n e a r X s u f f i c i e n t l y small p .

0o the unique s o l u t i o n X

= 0(u)

-17The phase portrait of Eq. (1.7) follows by considering the locally linearized system near X , of which the coefficient matrix A + p Df(X ) has eigenvalues which depend continuously on the perturbation parameter p. For small p, compared with |Re(A. _ ) | , (Re(A .) * 0) the eigenvalues of A + p Df(X ) cannot cross the is also a hyperbolic singularity. From Hartmanimaginary axis so that X

Grobman's theorem, then follows that the systems (1.6) and (1.7) have phase portraits near (0,0) which are topologically equivalent. Hence, phase portraits near hyperbolic points are structurally stable with respect to C1-perturbations, phase portraits near nonhyperbolic points may loose their topological structure if C' -perturbations are imposed on the dynamical system. This phenomenon where a small variation of the system causes a change of topological structure is called a bifurcation. If the changes are caused by perturbations containing parameters, those parameters which actually cause bifurcation are called bifurcation parameters. Two types of bifurcations in phase portraits may be distinguished: local^ bifurcations and global bifurcations. The former appear if the topology of the phase portrait is only .locally affected. Such bifurcations can be observed with local analyses, they occur in particular if nonhyperbolic points are present; bifurcation changes the topology of the phase portrait only in a small neighbourhood of the nonhyperbolic singularity. In those cases where local analyses fail to detect bifurcation effects, the bifurcation is called a global bifurcation. Saddle-connections and multiple limit cycles are well known examples of global bifurcations. Let us discuss them briefly for phase portraits in R* . Saddle_connection Consider a dynamical system in R* having a phase portrait which contains a special trajectory connecting two (hyperbolic) saddle points. Suppose that the system is perturbed such that the saddle connection breaks up resulting in its disappearance. Although the topology in the phase plane is significantly altered, see Fig. 1.6, the disappearance of the saddle connection cannot be detected by only a local examination of the phase portraits; to observe the broken saddle connection a broader view, containing the saddle points together with the separatrices, is necessary. The imposed perturbations have a so-called global effect on the trajectory pattern and as a result the transition process is called a global bifurcation.

-18-

saddle connection

broken saddle connection

Fig. 1.6. Perturbation o f saddle connection, global bifurcation.

A n o t h e r e x a m p l e o f a global bifurcation m a y appear i f a n isolated closed path: T is embedded i n a trajectory pattern in R' as shown i n F i g . 1 . 7 - A s s u m e that

for i n c r e a s i n g all tending to T

t t h e t r a j e c t o r i e s in the outer region are spiralling inwards, if t . I n the inner region o f T they tend to T for t >

and t o a focal p o i n t f o r t >. T h e isolated closed path T

is a (semistable) starting

limit c y c l e , w h i c h cannot b e reached i n finite time along trajectories in a neighbourhood o f T . c

The p h a s e portrait as sketched in Fig. 1.7 can b e generated by the p a t h s o f t h e system r = -r(r-l)2 , 9 = 1

where (r,8) are polar coordinates i n the phase plane. T h e closed path T found o n the c i r c l e r = 1.

is then

7^1
u=0 u>0 Fig. 1.7. Perturbation of limit cycle, global bifurcation.

Consider a perturbation of the system by adding the linear term ur (u 6 R) to the right-hand side of r; u appears to be a bifurcation parameter.

-19The phase portrait of the perturbed system r = -r(r-l)2 + ur, 9=1

has two closed paths: r. _ = 1 J~v if u < 0 and no closed path if p > 0. The appearance (disappearance) of closed paths has a global effect on the topological structure of the phase portraits which can not be established by only a local examination of the trajectory pattern. The previous remarks point out that bifurcations - local or global - may appear if the phase portrait is structural unstable. Nonhyperbolic singular points appear as elements in phase portraits that may cause local bifurcations; on the other hand saddle-connections and closed paths can give rise to bifurcations with a global character. Whether a bifurcation actually occurs is not only determined by the presence of structural unstable elements in a phase portrait but depends also on the class of perturbations that is admitted. The last statement applies also to global bifurcations, as can be shown by
3H 3H

perturbing a Hamiltonian system: x = , y = - global bifurcation to appear.

such that the perturbed system

remains Hamiltonian. Then closed trajectories can remain closed preventing a

-204. Higher-order singularities in R2 In p a r a g r a p h 3 we have seen t h a t hyperbolic and nonhyperbolic p o i n t s are structurally unstable with respect to an a p p r o p r i a t e c l a s s of p e r t u r b a t i o n s . T r a j e c t o r y p a t t e r n s near nonhyperbolic p o i n t s can be very complex and in general, as will be seen in the next paragraph, they will change i n t o p a t t e r n s near combinations of hyperbolic singular points if the system is perturbed. From t h i s point of view one might argue that nonhyperbolic points are far from i n t e r e s t i n g because they are so 'exotic' that they are hardly met in practical situations where small perturbations, always present, lead them desintegrate. Obviously, a d i f f e r e n t view results if non-hyperbolic points can be helpful to analyse transition processes in phase p o r t r a i t s , i . e . if an appropriate change of parameters in a dynamical system changes the topological structure. In that case, special combinations of parameter values e x i s t at which one or more s i n g u l a r p o i n t s i n the phase p a t t e r n become nonhyperbolic. For such a parameter combination the system i s called degenerate or nongeneric; however, a small change of these parameter values can convert the nongeneric system into a non-degenerate or generic one. Such a degenerate s t a t e of the system marks a point in the parameter space at which the topological properties of the system change qualitatively. These observations give us sufficient motivation for a more detailed treatment of these 'exotic' nonhyperbolic points. The discussion will be restricted to nonhyperbolic points in R2 since only these will be found to occur in the subsequent part of this work. The theory of nonhyperbolic singular points in R2 i s well developed by Andronov e t . al (1973) and the following review of the most important results serves as a b a s i s for applications. For a profound treatment including mathematical proofs, we refer to Andronov e t . al (1973) Andronov considers an i s o l a t e d nonhyperbolic s i n g u l a r point of an analytic vector field in R2 such that the expansion near the singular point involves at l e a s t one first-order term. Then a distinction can be made between singularities having one or both eigenvalues, zero.

-21Higher-order singular points with one zero eigenvalue; p * 0 Suppose that 0(0,0) is an isolated singular point of a planar system with one nonzero eigenvalue. Then the system can be written as x = P(x,y) = ax + by + P2(x,y) y = Q(x,y) = ex + dy + Q2<x,y) where P2(x,y) and Q_(x,y) are analytic with terms not lower than second degree and for the eigenvalues in 0(0,0) we have p=A. + A q = A.A =a+d * 0

= ad - be = 0

This system can be transformed by a nonsingular linear coordinate transformation in the canonical form x = P(x,y) = P2(x,y) y = Q(x,y) = y + Q2(x,y)

For isolated points i t i s assumed that P_(x,y) ^ 0. For this system Andronov has shown t h a t t r a j e c t o r i e s e x i s t t h a t tend to the s i n g u l a r point in a d e f i n i t e d i r e c t i o n (semipaths). These semipaths tend to 0(0,0) only in the directions 0, p, n and ^-; only one semipath e x i s t s in the d i r e c t i o n -z and only one in the direction ~-; these two special semipaths are denoted L1 and L_ respectively. To obtain the p o s s i b l e topological structures near the singular point consider the equation y + Q ? (x,y) = 0. By the implicit function theorem t h i s equation has e x a c t l y one s o l u t i o n y = <p(x) in a neighbourhood of 0 ( 0 , 0 ) , where <p(x) i s analytic and obeys the condition <p(o) = <t>' (o) = 0. The curve y = <p(x) i s an isocline of horizontal directions (y = 0) of the vector field. The next step i s to define a function:
*(x) = P 2 (x, <P(X))

-22-

which g i v e s t h e v a l u e of x on t h e c u r v e y = 0 . B e c a u s e 0 ( 0 , 0 ) i s (?2 ^ 0 ) , i|i(x) assumes t h e form *(x) = A x m + o ( x m ) , m > 2 m where A is the first nonvanishing coefficient (A

isolated

* 0) and m is integer.

Let us consider the neighbourhood U. of 0(0,0) bounded by a circle C with radius <5, see fig. 1.8. The curve y = <t>(x) d i v i d e s t h e domain U, i n t o two p a r t s . S i n c e y = y + Q 2 ( x , y ) , y > 0 i n t h e upper r e g i o n , s o t h a t v e c t o r f i e l d p o i n t s upwards t h e r e . the

*P(x)

S i m i l a r l y we have downwash i n t h e lower r e g i o n . T h i s implies t h a t t h e semipaths L. and L_, l y i n g above and below c u r v e y = <p(x) r e s p e c t i v e l y , s t a b l e . Since <t>(x) i s a n a l y t i c , y = <p(x) c o i n c i d e s with the the are uneither x-axis

(<p(x) = 0) or i t h a s , i n a s u f f i c i e n t l y s m a l l neighbourhood of 0 ( 0 , 0 ) , no o t h e r Fig. 1.8. Neighbourhood U- of 0(0,0) o p o i n t s than 0 ( 0 , 0 ) i n common with the xaxis.

In the first case the positive and negative x-axes are obviously semipaths of the system, while in the second case (<p(x)^0) the trajectories must cross the curve y = <P(x). These trajectories are passed along in a direction depending on the sign of t|>(x). With the expression i | i ( x ) sign A we observe that this depends on the A x + m m, which implies that four different cases have to and on the parity of > 0 or < 0, m is odd or even). According to Andronov, these

be considered (A

considerations lead to a classification of topological structures which is summarized in Andronov's theorem 65 (p. 3^0, Andronov et. al, (1973))Theorem 65. Higher-order singular points with only one zero eigenvalue Let 0(0,0) be an isolated singular point of the system x = P_(x,y) and y = y + Qp(x,y) where P_ and Q ? are analytic and have series expansions near 0(0,0) of which the lowest-order terms are at least quadratic.

-23Let y <p(x) be the solution of y + Q2(x,y) + 0 in the vicinity of 0(0,0) and

assume that the series expansion of the function *(x) = P 2 (x, <?(x)) has the form t | i ( x )= A x m where m 2 and A * 0. Then: m > 0, 0(0,0) is a topological node with an infinite

1. If m is odd and A

number of semipaths in the directions 0,n and exactly one semipath tending to 0 in the direction -= and also one in the direction ~-, 2. If m is odd and A < 0, 0(0,0) is a topological saddle point whose separa-

m _ t r i c e s (semipaths) approach 0(0,0) in the d i r e c t i o n 0, -~, n and ~ - , respectively. 3. If m i s even, 0(0,0) i s a saddle-node, i . e . a singular point whose neighbourhood c o n s i s t s of one p a r a b o l i c s e c t o r (nodal type) and two hyperbolic*) sectors (saddle-type). If A < 0 the hyperbolic sectors contain a segment of the p o s i t i v e x - a x i s , if A > 0 they contain a segment of the negative x-axis. By definition the order of a singularity i s equal to m. The full proof of this theorem may be found in Andronov e t . a l (1973) PP- 337 ff. Typical phase plane p i c t u r e s near these multiple-equilibrium points are shown in Fig. 1.9

^i

<

ktopological saddle saddle-node

topological node

Fig. 1.9. Higher-order singularities in R2 with one zero eigenvalue. Higher^order singular goints_having both_eigenvalues_zero_g_=_0 In t h i s section we consider the system x = ax + by + P 2 (x,y),
with the assumption p = a + d = 0 q = ad - be = 0 |a| + |b| + |c| + |d| * 0

y = ex + dy + Q 2 (x,y)

*) Other examples of hyperbolic sectors appear near a hyperbolic saddle point where four hyperbolic sectors can be distinguished.

-2kSuppose 0(0,0) is an isolated singular point at the origin and P,(x,y) and Q2(x,y) are analytic in the vicinity of 0. The series expansions of P (x,y) and Qp(x>y) involve terms not lower than second order. Then a nonsingular linear transformation exists (c.f. Andronov p. 3^7) bringing the system into the normal form x = y + P2(x,y) y = Q2(x,y)

The topological structure of 0(0,0) is found by considering the isocline of vertical directions: y = < t > x ) where <p(x) satisfies <p(x) + P-,(xt <p(x)) = 0. < p ( n On this curve the functions 3P*(x) = Q2(x, <p(x)) and p(x) = j ^ may be evaluated and expanded as i ( i ( x )=A x + m where A and b , p(x) = b x + ' ' n 3Q(x, <f(x)) + j ^ (x, <p(x)J

are the f i r s t nonvanishing coefficients in these expansions and

m and n are integer. The topological s t r u c t u r e of phase p o r t r a i t s in R2 near singular points with zero as a double eigenvalue is then established by Andronov's theorems 66 and 67 (PP. 356, 362). Theorem 66, 67. Higher-order singular points having both eigenvalues zero Let 0(0,0) be an i s o l a t e d singular point of the system x = y + P_(x,y), y = Q2(xy);
l e t

*( x ) e the solution of y + P ? (x, y) = 0 near 0(0,0) and

assume t h a t the s e r i e s expansions of the function 4 > (x) = Qp(x, <t>(x)) 3P2 3Q2 and p(x) = - (x, <p(x)) + - (x, <p(x)) have the respective forms ox oy *(x) = A x m + ..., A * 0 m m p(x) K V ' = b nx + . . . , b n* 0 .

-25-

1. Let the number m be odd, n = 2J + 1 (J 2 1) and A . = b 2 + 4(8+1)4 . Then i f A > 0 the s i n g u l a r p o i n t i s a t o p o l o g i c a l s a d d l e - p o i n t , A < 0 t h e following p o s s i b i l i t i e s occur: - a focus o r a c e n t e r i f b = 0; i f b * 0, n > g; i f b * 0 , n = C, A < 0 n n n - a topological node if b * 0, n is even, n < 2; if b * 0, n is even, n n n = e, A 0 - a s i n g u l a r p o i n t w i t h an e l l i p t i c region i f b b * 0, n i s odd, n = 2, A 0. 2 . Let the number m be even, m = 22, ( I 1 ) . Then the s i n g u l a r p o i n t i s - a saddle-node i f b * 0, n < n - a cusp if b = 0; if b * 0, n i . n n The order of the singularity is by definition equal to m. The proof of this theorem may be found in Andronov et. al (1973) PP- 365 ffSketches of the phase portraits are shown in Fig. 1.10. * 0, n = odd, n < g; i f but if

-26-

topological saddle A m >0 m = odd = 2l+1

'
bn=0 b n *0< ,X<0 X = b+4A m (H/)

y
^^

focus, center

< '

n=even ' X>0

topological node

n=odc)

elliptic and hyperbolic sector

cusp

saddle-node

x=y + P 2 (x,y),y=Q 2 (x,y) y|x=0 = 4 J ( x , = P, x


A

mxm p(x)rbnXn-

da
y x=0

F i g . 1.10. H i g h e r - o r d e r s i n g u l a r i t i e s i n R2 with two e i g e n v a l u e s z e r o .

-27" 5. Bifurcation of vector fields, unfoldings In this paragraph we study local b i f u r c a t i o n s as they occur i n v e c t o r defined as X = f (X), X 6 Rn, u e Rk fields

near s st r u c t u r a l l y u n s t a b l e s i n g u l a r p o i n t s . Consider a v e c t o r f i e l d in R

which depends on the k-dimensional parameter u = (p , u_, . . . . u, ) . The vector field f (X) i s assumed to be analytic. The term bifurcation introduced in paragraph 3 was o r i g i n a l l y used by Poincare t o d e s c r i b e the s p l i t t i n g behaviour of s t a t i o n a r y s o l u t i o n s : X = X of the dynamical system X = f (X). These solutions, which are r e p r e s e n t e d as s i n g u l a r p o i n t s in the phase space, can be found by s o l v i n g f (X ) = 0 for X as a p o o function of u i f the Jacobian Df (X ) has no zero eigenvalues. However, i f the u o Jacobian has a zero eigenvalue, occurring at some u, say u , the system X = f (X) is structurally unstable and several branches of the solution X = X (u) n+k can come together at (X ,u ) in R The parameter set u where the system looses its structural stability is called: bifurcation set, it can be viewed as dividing surfaces bordering domains in the parameter space where the system is non-degenerate (generic). Variations in the parameter space intersecting the set (u|u = u } cause a change of the topological structure of the phase portrait. Such changes are called bifurcations and the corresponding parameter values are the bifurcation values. A one-parameter family of systems with k-1 relations between the parameters u. , u_, .... u, is represented at a curve A in the k-dimensional parameter space. Assume that A intersects the bifurcation set, u , and that the intersection is c transversal. At the intersection, structural unstability occurs, but due to the transversality condition the degeneracy can be removed by any small displacement along A. The corresponding perturbation is called a generic perturbation. Variations in the parameter space that coincide with u correspond to a nongeneric perturbation.

-28In this paragraph we focus the attention on bifurcations of isolated singular points. As we have seen these bifurcations are referred to as local bifurcations , so that the vector field is studied near degenerate singular points and the bifurcating solutions are also found in the neighbourhood of these points. Let us start with a simple example of a dynamical system in R1 . Consider the one-dimensional 'vector' field:
x = f u

"* " x '

(1.8)

which depends on the scalar parameter u. Here Df (x) 3x2, and the only bifurcation point is (x ,u (0,0). It is P 0) and three difeasy to check that f (x) = 0 has one solution if u i 0 (x ferent solutions if u > 0 (x =0, / p ~ ) . A qualitative picture of these solutions is given in Fig. 1.11a which shows the branches of singular points in a (x,u) space. This figure is called the bifurcation diagram; it shows the loci of singular points as a function of the parameters. The phase space is one-dimensional and coincides with the x-axis. The point x = 0 is stable for p S 0 and it becomes unstable for u > 0. The phase portraits which can occur after bifurcation are shown in Fig. 1.11b.

(b)
(jO:

bifurcation point

U>0:

-/Ji

Fig. 1.11. Bifurcation of f (x) = ux - x3. (a) bifurcation diagram, (b) phase portraits. The previous example of a one-parameter bifurcation gives the result of a oneparameter variation on the structurally unstable 'vector' field x = -x3. The perturbed system x = +ux - x' shows some possible bifurcations and is called an unfolding of the unperturbed system x = -x3 . Now the important question arises: is the unfolding given by eq. (1.8) a particular one, showing only some bifurcations, or has the unfolding a more general character so that all possible bifurcations of x = -x', within a certain class of perturbations (for example

-29analytic), are described. If the unfolding describes all possible bifurcations of the degenerate singular point, then it is called a universal unfolding. Hence a universal unfolding of a degenerate vector field is a family of bifurcating solutions which contains the bifurcation in a persistent way. It has the important property that it describes the bifurcation with the smallest number of parameters. This number is called the codimension of the degenerate vector field. The codimension of a bifurcation is the smallest dimension of the parameter space which contains the bifurcation in all its aspects. A universal unfolding is not necessarily unique since a coordinate transformation can lead to a differently formulated unfolding having the same dimension. Therefore the term 'universal unfolding' is not generally accepted in the literature; for example Shirer and Wells (1983) prefer the term versal unfolding to point out the nonuniqueness of the unfolding.

Bearing this in mind we return to the system x = -x' in order to find the universal unfolding of this system. Eq. (1.8) gives an unfolding with one parameter, but does it describe the bifurcation in all its aspects or are more parameters necessary for a universal unfolding? If the function f(x) = -x1 is subjected to a small perturbation f(x) f (x), it is obvious that f (x) possesses one, two or three zeroes near x = 0. However, the unfolding (1.8) gives either one (u < 0) or three (u > 0) zeroes, which indicates that (1.8) is probably not a universal unfolding of x = -x1 . Because it is not possible to introduce more than three zeroes locally, all these behaviours can be captured by the addition of the lower-order terms u- + u_x, so that a universal unfolding of x = -x' is represented by the twoparameter family fu(x) = jix + u2x - x' *) Equating f (x) to zero gives the loci of singular points x with these loci in the x ,p1,u_ space. (1.9) as a function of the

parameters u. and u_. Figure 1.12 shows the corresponding bifurcation diagram

) Other universal unfoldings such as f (x) = u.x + u2x2 - x* or f (x) = u. + u_x2 - x' have the same bifurcation characteristics: they can be transformed into eq. (1.9) by a suitable translation of the x-coordinate.

-30-

A s i n g u l a r point becomes degenerate i f Df(x ) has eigenvalues with zero r e a l part, thus if u_ - 3xz = 0. Eliminating x from this equation and the equation f (x ) = 0 we find the parameter combinations a t which bifurcation takes place. These parameter combinations are called the bifurcation set which s a t i s f i e s : Fig. 1.12. Bifurcation diagram of f (x)
y

+ P X

"

This s e t c o n s i s t s of two branches if u_ > 0, the corresponding one-dimensional phase p o r t r a i t s are shown in Figure 1.13. W e observe from the phase p o r t r a i t s that on a branch (u_ > 0) the nonhyperbolic point i s the one-dimensional variant of a saddle-node. Crossing a branch transversally implies the bifurcation of the saddle-node. Such a crossing i s in fact described by one parameter so that the branches represent a codimension-one bifurcation. At the point (u..,u_) = (0,0) where both branches terminate with a common tangent we have a nonhyperbolic onedimensional node; i t s bifurcation i s described by two parameters implying t h a t the point (u.. ,u_) = (0,0) represents a codimension-two bifurcation. This example of a codimension-two bifurcation i l l u s t r a t e s that the unfolding of x = - x ' given by eq. (1.9) c o n t a i n s , besides the generic bifurcation (already given by eq. ( 1 . 8 ) ) , also n o n g e n e r i c b i f u r c a t i o n s . The p o s s i b i l i t y to describe a l s o these nongeneric b i f u r c a tions i s realized by introducing the two parameter unfolding: f (x) = u 1 +y-x-x* , implying that the universal unfolding of x = -x' i s defined with two parameters involving a codimension-two bifurcation. If only generic b i f u r c a t i o n s a r e of i n t e r e s t i t i s enough to consider the Fig. 1.13. Bifurcation set and phase one-parameter unfolding as given by eq. p o r t r a i t s of x=u- +p_x-x' . (1.8).

{#-(#"

-31-

The preceding consideration about bifurcations of a one-dimensional system illustrates that concepts like unfolding and codimension can be valuable tools in bifurcation analyses. Especially if the aim is to develop a general theory about degeneracies and their universal unfoldings, the concept codimension plays an important role. It offers the possibility to develop a strategy for the study of these degeneracies whereby one starts by investigating codimension-one bifurcation, then codimension-two bifurcations etc. Such a strategy can ultimately lead to a classification scheme of bifurcations based on general considerations such as number of parameters, dimension of the phase space and constraints which account for the class of perturbations, see for example Shirer and Wells (1983). In the next table we have summarized a few examples of elementary universal unfoldings of degenerate singularities occurring in one- and two-dimensional systems.

One-dimensional universal unfoldings

form

codimension 1

unfolding

name fold

U 1 + u2x u ? x + p x< P x + p2x + p3x Two-dimensional universal unfoldings form


xy x2
- y2 x'
J

cusp swallowtail + p^x


3

butterfly

codimension

unfolding P j + p,,y + xy U2 + p . y + x2 - j | p x + p 3 x + p^y +

name

hyperbolic-umbilic

elliptic-umbilic
P2
+

y2

As we have mentioned before, the presented form of the unfoldings i s not necess a r i l y unique; s e v e r a l a l t e r a t i o n s can be found which account for the same b i f u r c a t i o n behaviour. These alterations follow from suitable coordinate transformations and rearrangement of the parameters. The l o c i of the nearby s i n g u l a r i t i e s : x (p , p p . ) , forming the bifurcation diagram can be portrayed as surfaces (boundaries) in the x - p-space (p e R ) . The complex geometrical structure of these surfaces near the origin

/ -32can be associated with well known elementary singularities appearing in c a t a s trophe theory. This correspondence with elementary catastrophes i s reflected by the names l i s t e d in the table.

-336. Center manifolds Before proceeding with the applications in fluid flow problems, there is a general technique which has to be discussed first, as it can simplify the analysis of bifurcation problems considerably. This technique has the effect of introducing coordinate systems in which computations are more easily carried out. After using it one is left with a reduced system of differential equations containing all the qualitative features of the bifurcation. It must be emphasized that this technique, called center manifold theory, has a local character and is only applicable to bifurcations of singular points. The center manifold theory provides a means for systematically reducing the dimension of the state spaces which need to be considered when analyzing bifurcations of a given type. Suppose we have a system of nonlinear ordinary differential equations: X = f(X) X e Rn

for which f(0) = 0. The linearized system at X = 0 may be written as X = Df(0).X If Df(0) has no eigenvalues with zero r e a l p a r t , then Hartman-Grobman's theorem s t a t e s t h a t the e i g e n v a l u e s , with positive and negative r e a l p a r t s , determine the local phase p o r t r a i t near X = 0. If t h e r e are e i g e n v a l u e s with zero r e a l p a r t s the t o p o l o g i c a l s t r u c t u r e can be quite complicated. Some examples in R * have already been given in t h i s chapter (paragraph ^ ) . A solution of t h e l i n e a r i z e d system, satisfying the i n i t i a l condition

X(t ) = X , X eR v o o o

i s the vector valued function

X(t;X o ) = exp(t Df(0)) XQ A general solution of X = Df(0) X i s obtained by l i n e a r s u p e r p o s i t i o n of n l i n e a r l y independent solutions: ( V f t ) ,


n 1 J X(t) = X C. J V (t) j=l

v (t)}

-3*where C. (j = 1, ... n) is a constant. A set of linearly independent solutions can be obtained from the real and imaginary parts of the vector valued functions exp(A.t) XJ where X"' are the (generalized) eigenvectors associated with the eigenvalues (real or complex) Aj(j = 1, ... n ) . Various subspaces spanned by (generalized) eigenvectors and filled with solution curves may be distinguished. n Xs

the stable subspace,

E : spanned by X'

n the unstable subspace, E : spanned by X1 ... X n c c E : spanned by X' ... X

the center subspace,

n s where X1 , . . . X are n (generalized) eigenvectors of which the eigenvalues s s s have negative real parts, n X1 , ... X are n (generalized) eigenvectors of which the eigenvalues have positive real parts and n X1 , ... X are n (generalized) eigenvectors with eigenvalues having zero real parts. Since the dimension of the whole phase space is n we have n + n + n = n. s u e The adjectives s t a b l e , unstable and center r e f l e c t the behaviour of solutions in t h e i r eigenspaces respectively. a r e c h a r a c t e r i z e d by exponential decay (monotonie or u c o s c i l l a t o r y ) , t h o s e lying in E by exponential growth and those lying in E by neither. In Fig. 1.14 we show some examples of invariant subspaces in R2 and R' . s u For nonlinear systems one can define the invariant subspaces W and W being the s u s u nonlinear analogues of E and E r e s p e c t i v e l y . The subspaces W and W a r e
g

Solutions lying i n E

frequently denoted as stable and unstable manifolds, they are tangent to E and E u at the singular point where f(X) = 0. The existence of these manifolds for nonlinear systems is established by the following theorem.

-35Df(0)=(2
u 0l

E = span{(1,OI,<0,1)}

(b)

Df(0) = (J.)
E u = (1,0) E s = (0,1) Ec = 0

co

Df(o,

=(- , o:.?)

EU=0,ES=(O.O,1) E c = span{(1,0,0),(0,1,0)}

Fig.

1.14. Subspaces of linear flows.

Stable Manifold Theorem for hyperbolic singular points Suppose that X = f (X) h a s a h y p e r b o l i c singular point X , then there exist stable and unstable manifolds W (X ) , W (X ) , of the same dimensions n , n as o' o s u those of the eigenspaces E , E the function f. The proof of this theorem has been given by Hartman (1964) and more recently by Carr (1981). We note that this theorem says nothing about the existence of center manifolds c r W for nonlinear systems. For the case that the nonlinear system is C the existence of center manifolds is established in the center manifold theorem which includes the results of the stable manifold theorem. The first proof of the center manifold theorem has been given by Kelley (1967). of the linearized system; at X , W (X ) is tangent to E s and WU(X ) is tangent to E u ; WS(X ) and WU(X ) are as smooth as

-36Center Manifold Theorem Let f be a C v e c t o r f i e l d on R vanishing a t X = X so t h a t f(X ) = 0. o o The s p e c t r u m of e i g e n v a l u e s AA of Df(X ) i s divided i n t o t h r e e p a r t s , a , o , a with s u c A e as A a A6 a c u if if if Re A < 0 Re A = 0 Re A > 0 and E u r e s p e c t i v e l y . Then at a t X . The

The corresponding g e n e r a l i z e d eigenspaces a r e E , E

t h e r e e x i s t C s t a b l e and u n s t a b l e manifolds W and W tangent t o E and E X ; r e s p e c t i v e l y and a C c e n t e r manifold W t a n g e n t to E s u c manifolds W , W and W a r e each f i l l e d with s o l u t i o n c u r v e s .

The c e n t e r manifold theorem implies t h a t a n o n l i n e a r s y s t e m f (X) can be s p l i t s+u c " u p , n e a r a s i n g u l a r p o i n t X , i n t o two s u b s y s t e m s f and f of which the l i n e a r p a r t s have e i g e n v a l u e s respectively. Then one may w r i t e the system X = f(X) near the s i n g u l a r p o i n t X : (X ,Y ) mally as X = f C (X,Y) Y = f
S+U

w i t h n o n z e r o and w i t h z e r o r e a l

parts, for-

= DfC(X ,Y ).(X-X ) o o o
S+U

+ F(X,Y) (1.10)

(X,Y) = Df

(X ,Y ).(Y-Y ) + G(X,Y)
0 - 0

n n +n c s u where X R , Y e R and F, G are nonlinear higher-order terms which vanish at (X o .Y o ). Since the c e n t e r manifold i s t a n g e n t to E plicitely Y-Y = h(X) with h(X ) = Dh(X ) = 0 o o the at (X ,Y ) i t can be w r i t t e n ex-

The c e n t e r manifold W may be viewed a s a p a r t i c u l a r s e t of s o l u t i o n s o f

o r i g i n a l s y s t e m X=f (X) , s a t i s f y i n g the a d d i t i o n a l c o n s t r a i n t s f(X )=Df(X )=0.

-37They may be calculated by substituting Y = Y in Dh(X) [DfC(Xo.Yo)(X-Xo)


+

+ h(X) into eq. (1.10), resulting

F(X.YQ+h(X))] (1.11)

-DfS+U(XQ,Yo)h(X) - G(X,Y +h(X)) = 0

with the boundary conditions h(X ) = Dh(X ) = 0. Equation (1.11) i s in general a p a r t i a l differential equation which cannot be solved exactly in most cases, but i t s solution can sometimes be approximated by a s e r i e s expansion near X . If such an approximate s o l u t i o n i s found, the original system may be p r o j e c t e d on the c e n t e r manifold r e s u l t i n g i n t o the following reduced system X = DfC(X ,Y ).(X-X ) + F(X,Y +h(X)) o o o o
Since h(X) is tangent to E

(1.12)

at the singular point, the solutions of the reduced

system provide an approximation of (1.10) near X . This conclusion was obtained, independently, by Henry (1981) and Carr (1981) when they proved that the properties of (1.10) near (X ,Y ) and of (1.12) near X are the same. It implies that the bifurcation properties of a degenerate singularity can be obtained by analysing the nonlinear system on its center manifold. Although this method looks straightforward, its succes depends largely on the possibility whether a center manifold can be found or at least approximated in a suitable way.

Other difficulties that can appear are the non-uniqueness and loss of smoothness of s o l u t i o n s in the center manifold. The boundary conditions h(X ) = Dh(X ) = 0 are not sufficient for a unique solution of eq. (1.11). In the following examples we point out some of these aspects ( p o s s i b i l i t i e s , limitations) of the use of center manifolds. In the f i r s t example c e n t e r manifold theory will be used in order to find the phase p o r t r a i t s of the two-dimensional system x =
EX

+ xy - x'

y = -y + yJ - x2
if the real-valued parameter e varies near zero.

-38-

If e = 0 the system has a degenerate s i n g u l a r point a t the o r i g i n ( 0 , 0 ) ; the corresponding c o e f f i c i e n t matrix of the linear part i s given by L _1 . The eigenspaces are: E s = (0,1), E c = (1,0), Eu = *. Using theorem 65 (paragraph 4) the singularity at the origin is a stable topol o g i c a l node; an i n f i n i t e number of paths approach the node along t h e xdirection,
To obtain the bifurcation solutions for c * o we consider the center manifold of the suspended system in R1 x = ex + xy - x' 'c = 0
y = - y + y* - x 1

(1.13)

where the 'parameter' e i s seen as ' t r i v i a l ' dependent variable. The coefficient matrix of the linear part becomes for c = 0

0 0 0

0 0

0^ 0 0-1 with eigenvalues A = A = 0, A

The corresponding eigenspaces are E C = (1,0,0) x (0,1,0), E s = (0,0,1) and E =. Now we seek a center manifold h = h(x,e) tangent to E , which is a solution of eq. (1.11), satisfying the boundary conditions h(0,0) = |^ (0,0) = |^ (0,0) = 0. Equation (1.11) becomes in this case 3h 9h\ icx - x' + xhi . ., , _ I, I + h - h2 + x' = 0 l 3x 3e' l o ' The center manifold will be approximated by h(x,e) = eye2 + a2xc + a,e* + ^x' + p2x2e + p xe2 + p^c1 + 0(4)

where 0(4) denotes terms of 0[(|x| + |e|)*].

-39-

Fig. 1.15. Center manifold W : h(x,e) = -x2 + 2x 2 e + 0(4). Equating powers of x ! , x t , . , . E ' we find for the center manifold {see Fig. 1.15) W : h(x,e) = -x2 + 2x2 e + 0(4) The reduced system which governs the b i f u r c a t i o n caused by c, follows as the projection of system (1.13) on W : ex - 2x3 + 2ex3 + 0(4), E=0

The vector f i e l d on W i s a one-parameter unfolding of the one-dimensional degeneracy: -2x'. The unfolding has a l s o a s i n g u l a r i t y a t x = 0 and two s t a b l e neighbouring singular points at x = ff* + 0 ( e ' ) but only for e > 0. The singular point x = 0 appears to be s t a b l e for E < 0 and unstable for e > 0. The behaviour of the reduced system on W enables us to draw the trajectories in the x,y-phase plane i f e v a r i e s near z e r o . Since the unstable eigenspace Eu appeared to be empty, the trajectories tend to W for increasing t . see Fig. 1.16.

-40-

(a): C<0

(b): E>0

Fig. 1.16. Bifurcation of: x = ex - x' + xy, y = -y + y' - x2 . The next example illustrates that a center manifold is not always unique. Consider the two-dimensional system x = -x' , y = -y

which has a isolated singular point (topological node) at the origin. The center c c eigenspace E is formed by (1,0) and the center manifold W : h(X) is a solution of -x'h' + h = 0 with h(o) = h'(o) = 0

The general solution of this differential equation is h(x) = C.exp(- ) Since the solution satisfies h(o) = h'(o) = 0 irrespective the value of C, it must be concluded that an infinite number of center manifolds exists. Since all derivatives of exp(- ?) vanish at x = 0 only the center manifold

h - 0 is analytic, and all other center manifolds are C .

-41-

7. An approach to physical unfoldings in flow patterns As announced in the introduction, this thesis deals with bifurcations in v e c t o r f i e l d s as they arise as velocity fields in gas- and fluid motions; e.g. conical flows (Chapter II) and viscous flows retarded near a smooth s u r f a c e (Chapter I I I ) . In general a t s t a g n a t i o n p o i n t s these vector fields will have isolated s i n g u l a r i t i e s which can degenerate under c e r t a i n c o n d i t i o n s . To o b t a i n a u n i v e r s a l unfolding of such a degeneracy one can follow a mathematical and a physical approach. The former, being very s t r a i g h t forward, determines the u n i v e r s a l unfolding of a degenerate s t a t e on the b a s i s of i t s topological features such as codlmension, center manifold and on the b a s i s of the c l a s s of perturbations admitted. In t h i s way, the complete bifurcation can be described with the smallest number of parameters (say u.. , . . . u, , k e R) and general unfolding principles can be used. Although t h i s approach i s very a t t r a c t i v e i t s application involves problems of different kind. F i r s t , degeneracies occurring in flow fields appear scarcely in those convenient forms being necessary for a straightforward a p p l i c a t i o n of unfolding theory. These degeneracies - t o g e t h e r with i t s unfoldings - are more or less of particular type because the corresponding degenerate solutions have to s a t i s f y the p a r t i a l d i f f e r e n t i a l equations which govern the physics of the flow ( e . g . conservation of mass, momentum and energy). Consequently, i t means t h a t we are l e s s i n t e r e s t e d in finding the universal unfolding of the degeneracy but more in those unfoldings which a r e allowed by the flow equations. Let us c a l l such a permitted unfolding a physical unfolding. A second difficulty stems from the problem of i n t e r p r e t i n g the involved p a r a meters of the universal unfolding. The physical interpretation of these parameters i s a necessary but by no means easy task, as the physically relevant parameters may be related to the bifurcation parameters u1 , . . . a in a very complicated way. Furthermore, the number of r e l e v a n t physical parameters may exceed the number of bifurcation parameters so that a single b i f u r c a t i o n parameter may be c o n s t i t u t e d by a combination of s e v e r a l physical parameters. To find these combinations, one needs additional information ( e . g . from the flow problem) which cannot be d e r i v e d from the universal unfolding.

-l\2On the other hand, if the number of relevant physical parameters is smaller than the codimension, it implies the existence of a dependency between the bifurcation parameters. This dependency can significantly reduce the possible bifurcation solutions with respect to those given by the universal unfolding. Moreover if the number of physical parameters is not known beforehand there is a risk that unallowed bifurcations will creep in, which remain unnoticed, and lead to a distorted view about transition behaviour in flow patterns. In a physical approach, the above mentioned difficulties are circumvented since a degenerate state in the flow pattern and the corresponding physical unfolding appear simultaneously if the governing flow equations are evaluated near singular points. Actually, the evaluation results into approximate solutions which satisfy these equations to a certain order near the singularity. These approximations contain several unknowns which cannot be determined in a local analysis. For specified values of the unknowns the singularity attains a degenerate state indicating that the involved unknowns can be viewed as bifurcation parameters. Then the physical unfolding of the degenerate singularity follows quite easily by varying the involved parameters with respect to bifurcation values at which the degenerate singularity was found. The author is inclined to favour the physical approach as long as the emphasis lies on local bifurcations in vector fields that are generated by physical systems.

-i3-

8. References Andronov, A.A., Leontovich, E.A., Gordon, I.I. and Maier, A.G. (1973) Qualitative theory of second-order dynamical systems. Wiley, New York. Coddington, E.A. and Levinson, L. (1955) Theory of ordinary differential equations. McGraw-Hill, New York. Carr, J. (1981) Applications of center manifold theory. Springer-Verlag, New York, Berlin. Guckenheimer, J. and Holmes, P.J. (1983) Nonlinear oscillations, dynamical systems and bifurcation of vector fields. Springer-Verlag, New York, Berlin, Heidelberg, Tokyo. Hartman, P. (1964) Ordinary differential equations. Wiley, New York. Henry, D. (1981) Geometric theory of semilinear parabolic equations. Lecture notes in mathematics, Vol. 840. Springer-Verlag, New York, Heidelberg, Berlin. Jordan, D.W. and Smith, P. (1977) Nonlinear ordinary differential equations. Clarendon Press, Oxford. Kelley, A. (1976) The stable, center s t a b l e , c e n t e r , c e n t e r u n s t a b l e and u n s t a b l e manifolds. J. Diff. Eqns, 3, 5**6-570. Reyn, J.W. (1964) C l a s s i f i c a t i o n and d e s c r i p t i o n of the s i n g u l a r p o i n t s of a system of three linear differential equations. J. Appl. Math. Phys., 15, 5^0-557 Shirer, H.N. and Wells, R. (1983) Mathematical structure of the s i n g u l a r i t i e s a t the t r a n s i t i o n s between steady s t a t e s in hydrodynamic systems. Lecture notes in physics. Vol. I85. SpringerVerlag, Berlin, Heidelberg, New York, Tokyo.

-45-

Chapter II

Topology of conical flow patterns

1. Introduction
1^1. Concepts and definitions The qualitative theory of dynamical systems will be applied to three-dimensional inviscid flows with conical symmetry. Such flows, which are called conical flows, have the specific property that the velocity and the quantities defining the state of the gas, e.g. pressure and temperature, are constant along rays emanating from a common point in the physical space. This point is called the center of the conical field. Inviscid conical gas flows embody an important class of flows around conical configurations of practical interest, such as delta wings having arbitrary cross-section, inlet configurations, nose cones, wing-body junctions and internal flows in nozzles and diffusors. Due to the conical symmetry, these flows can be described by functions of two independent variables defining the position of the rays. The structure of a conical flow may be represented on a unit sphere centered at the conical center. The velocity vector V may be decomposed into a radial component V unit sphere and a component V normal to the tangent to it; the latter defines a vector field

on this sphere. Integration of this vector field yields lines on the unit sphere which will be called conical streamlines. The physical significance of a conical streamline may be explained quite easily. Consider the spatial streamlines passing through a common ray. Since the flow is assumed to be conical these streamlines form a conical streamsurface, which is built up by sequential rays emanating from the conical center. Taking the intersection of a conical stream surface with the unit sphere, a conical streamline, is obtained. Similarly, conical streamlines can be defined on a plane surface Z at unit distance from the conical center. Such a plane is usually called a cross flow plane and the conical streamlines follow as the intersections of conical streamsurfaces with Z. Analogous to the situation on the unit sphere, conical streamlines in the cross flow plane Z can be taken as integral curves of a vector field in Z. This vector field may be obtained if the velocity vector V is decomposed nonorthogonally into a radial component and into a component in the cross flow plane. The latter, which is called cross flow velocity, yields the vector field in the cross flow plane.

-46Points in the cross flow plane and on the unit sphere where the cross flow velocity vanishes, thus the velocity has only a radial component, are called conical stagnation points. These points the vector field are a critical points where the conical streamline direction is undetermined. l;2i_A_brief_survey_of_conical flow.theory The early theoretical work on conical flows is largely based on linearized theory and originates from the work of Busemann (1929) on circular cones in supersonic flow. From that time linear theory of conical flows is developed and applied to configurations of aerodynamic interest by several authors. Much of the early work culminated in the studies of Germain (1949) and Goldstein & Ward (1950). The development of a nonlinear theory for conical flows starts with the work of Taylor & Maccoll (1933) when they studied the axisymmetric flow around a circular cone. A great variety of theoretical aspects of nonlinear conical flows is treated by Bulakh in the late fifties. He concentrated his efforts on the formulation of boundary value problems and on finding approximate solutions for them. These solutions are discussed by Bulakh with special emphasis to their properties and singular behaviour, in particular on the Mach-cone and near shock waves. A further study on the properties of nonlinear isentropic conical flows is performed by Reyn (i960), where he studied the solution of the nonlinear partial differential equation from the point of view of differential geometry. In doing so Reyn commented and clarified and Fowell (1956). All these investigations contributed significantly to the theory of nonlinear conical flows which, to a certain extent, has reached a final form by the publication (in Russian) in 1970 of Bulakh's book: 'Nonlinear Conical Gas Flows'. Ever since that time the main attention diverts to the application of numerical methods to obtain solutions for flows around cones and delta wings at incidence. Although these methods give accurate solutions at moderate incidence, for larger angles of attack two fundamental difficulties arise which prevent a straightforward numerical solution. First, as the angle of attack is increased, the cross flow velocity will increase to supersonic values in some parts of the flow field. This changes the nature of the governing partial differential equations from elliptic to a mixed elliptic-hyperbolic type and results in the appearence of a conical supersonic region terminated by an embedded shock wave. certain discrepancies, occurring in the flow around a flat delta wing with supersonic leading edges as given by Maslen (1952)

-klAnother difficulty is that some flow properties such as entropy, density and radial velocity can become multivalued at conical stagnation points. Since entropy gradients can be identified with rotationality of the flow (Crocco's theorem), these conical stagnation points can appear as vortical singularities. The presence of vortical singularities, which was a severe obstacle for a straightforward numerical calculation, see for example Stocker and Mauger (1962), did revive the attention to nonlinear conical flow theory with special emphasis on the conical flow structure near conical stagnation points. It is of value therefore, to evaluate as systematically as possible by means of a local analysis, the possible flow structures near such points, so that in a particular flow problem the qualitative basis for a numerical procedure can be selected with more certainty. In conical flows with entropy gradients the entropy remains constant on conical streamlines. Then, if in a conical stagnation point various streamlines merge, a vortical singularity is formed. This conjecture was put forward by Ferri (1951) when he discussed the supersonic flow past a circular cone at incidence. Since the appearance of Ferri's paper investigations of the flow near conical stagnation points show an emphasis of interest in the possible conical streamline patterns together with the related pressure distributions near such points. Melnik (1967) constructed some approximate solutions of the nonlinear inviscid conical flow equations near conical stagnation points attached to a body surface. These solutions involve entropy gradients in the flow. When the streamline pattern is related to the corresponding pressure distribution on the body surface no unique correspondence was found. Bakker (1977) showed that for these solutions a unique correspondence may be obtained if the pressure distribution normal to the body surface is also taken into account. Both investigations indicate that the presence of entropy gradients does not affect the topological structure of the conical streamline pattern that corresponds to a given pressure distribution. This result was further confirmed in the special case of the conical stagnation points in the flow past slender circular cones at high incidence, as calculated using slender body theory (Smith 1972), or linearized theory (Bakker & Bannink 1971*).

It is of interest, therefore, to make a further study of the topological structure of conical flows near conical stagnation points using the assumption of potential flow. An advantage of this approach is that the nonlinear conical flow equations reduce to a single second-order equation for the conical potential for which solutions are simpler to obtain. Moreover in a conical stagnation point this equation becomes Laplace's equation which is also satisfied by the velocity

-48-

p o t e n t i a l i n incompressible plane flow. Stagnation-point solutions for incompressible plane flows a r e then used as a guide to conical s t a g n a t i o n p o i n t solutions. Since conical stagnation points appear as c r i t i c a l points in a vector field in R2 the qualitative theory of dynamical systems, as o u t l i n e d i n the previous chapter, appears to be particularly useful. l 1 2;_52Di5i_5treamlines 1 _conical_stagfnation_Doints W e w i l l d i s c u s s some examples of conical flows and consider the corresponding topological structures in the cross flow plane S. Let the flow be described by a righthanded c a r t e s i a n coordinate system x,y,z and l e t the velocity vector V be decomposed i n t o i t s components u,v,w along the x , ' y and z axes, respectively. The c o n i c a l center will be taken a t the origin (0,0,0) and the cross flow plane 1 i s chosen normal to the x-axis. Since the flow i s conical i t i s advantageous to use the c o n i c a l c o o r d i n a t e s n = y / x , t, = z/x t o describe the cross flow pattern in the plane I. If the v e l o c i t y vector V = u i + v j + w k i s decomposed nonorthogonally into a component g along the ray r = x i + y j + z k and a component g i n the plane X, then the l a t t e r defines the cross flow velocity vector. The direction of q_ may be obtained as follows. Because q_ l i e s in I i t can be written as g = o.i + v j +w k

similarly the component along the ray is given by

lsrl q = . (x i + y j + z k) "r |r |
The cross flow components v and w follow from the requirement '
>

Y = 92 + 3 r
which yields

l9rl

= u/x

k
v =v -

IsJ r'

y = v - uu

k |

-49w_ = w 2 ' 3 r' z = w - UC

k |

This implies that the conical streamlines in T are solutions of the equation dn _ v - uu dC w - uC or of the equivalent system
n

= V j = v - un,

C = w z = w - uQ

(2.1)

where the dot represents differentiation with respect to a parameter along the streamlines. Equations (2.1) will be considered as a dynamical system in R2 whose phase trajectories coincide with the conical streamlines in the cross flow plane. Conical stagnation points occur on those rays where the velocity vector is aligned with a local ray, yielding g =0 or v - un = w - uC = 0

Conical stagnation points may be identified with the singular points of system (2.1) ; they serve as the indispensable flow elements which may be used to compose more complicated conical flow patterns. In that way the knowledge about different types of conical stagnation points contribute to the understanding of the topology of conical flow patterns of increasing complexity. To obtain a certain familiarity with the characteristic features of conical flow topology we proceed by discussing first the most simple example of a conical flow: the uniform parallel flow. Consider a uniform parallel flow with a constant velocity u in the positive xdirection. This flow will be considered in the cross flow plane I. Since u = u^, v = w = 0 the conical streamlines in I may be obtained from Eq. (2.1) as n = -u n; C = -u C This system has the general solution !; = const, n. The conical s t r e a m l i n e s are s t r a i g h t l i n e s p o i n t i n g to the o r i g i n where a conical s t a g n a t i o n p o i n t i s formed, see Fig. 2 . 1 .

-50-

conical center

I . / ^ cross flow plane (x=const\)

Fig. 2 . 1 . Conical view of a supersonic flow u = U . For increasing time the conical streamlines are passed along i n a d i r e c t i o n indicated by the arrows if u > 0; for u < 0 this direction reverses.
> co
*

co

The trace p and the Jacobian q are given by p = -2ua and q = u2 indicating that the uniform flow parallel to the x-axis is represented as a starlike node in the cross flow plane I. For u
CO

> 0 we refer to it as a conical sink, and for u


CO

< 0

as a conical source. If the uniform flow i s not aligned with the x - a x i s the s t a r l i k e node may be found o u t s i d e the o r i g i n . Since the starlike node i t s e l f remains unaltered i t indicates that a uniform flow has a unique conical flow p a t t e r n without any regard to the s p e c i f i c o r i e n t a t i o n of the cross flow plane X. This conclusion may be i l l u s t r a t e d as follows. Assume that a uniform flow, defined by | v | = c o n s t a n t , has v e l o c i t y components u , v and w in the x, y and z-direction,
03 CD CO

respectively. The conical streamlines satisfy the equation u n u C

having the solution (c - ) = const, (n - )

The conical streamlines form a bundle of straight lines in 2 which point to a v w


CO OS

conical stagnation point at f t= , C = .


CO CO

The conical stagnation point is a starlike node (p = -2u , q = u 2 ), its location (, ] coincides with the intersection of V with 1. K
u

u '

-51We have seen, irrespective of the orientation of the cross flow plane, that the uniform flow in the cross flow plane is uniquely represented by a starlike node. However the opposite that any starlike node in 1 reflects a uniform flow is in general not correct. This may be verified easily if one considers the example of an axisymmetric flow where every plane through the x-axis is a streamsurface. Since the velocity normal to the streamsurface is zero, the conical streamlines must be straight lines forming a starlike node at the origin (0,0). l;^1_Transition_phenomenae in conical_flow_gatterns Consider a circular cone with semi apex angle 8 submerged, at incidence a, in a supersonic flow with free stream Mach number M . The semi apex angle, incidence and free stream Mach number are such that a conical bow shock originates from the apex of the cone, see Fig. 2.2. The flow field will be described in a right handed cartesian coordinate system x,y,z such that the x-axis coincides with the cone axis and the flow field is symmetrical with respect to the x-z-plane. The cross flow plane 2 will be chosen normal to the x-axis at unit distance from the cone apex {conical center).

Fig. 2.2. Circular cone at incidence in supersonic flow with attached bow shock. The boundary condition on the cone surface reads (v - un) + (w - uq) C = 0

T I

For a = 0 the flow is axisymmetric having a bow wave with circular cross section in 2. The conical streamlines outside and inside the bow wave are straight lines pointing to the origin (0,0).

-52-

Outside the bow wave the flow is uniform and inside of it we have a TaylorMaccoll flow. Since the flow is axisymmetric with no velocity component normal to planes through the x-axis the cross flow components obey the ( v - u n ) <;- ( w - u C ) relation

n = 0. The boundary condition implies that on the cone =

surface both equations can be satisfied only if the cross flow vector q

(v - un, w - u O vanishes identically. As a consequence the cone surface is the union of conical stagnation points which may be called a conical stagnation line. All conical streamlines terminate perpendicularly at the conical stagnation line, see Fig. 2.3a.

If the incidence is slightly increased the topology of the flow pattern alters drastically; the conical streamlines encompass the cone surface and terminate at the leeside in a common point, see Fig. 2.3. At moderate incidence (say a/8 <

1) and irrespective of the influence of viscosity, two conical stagnation points may be observed in the symmetry plane, one at the windward and another at the leeward generator of the cone. The conical stagnation point at the windward side is a saddle point, whereas at the leeward side a nodal point (vortical singularity) appears.

increasing incidence

(a)

(b) viscous

(c)
?

(d)

flow

,'

(e)

(f)

Fig. 2.3- Conical flow patterns for circular cone at incidence.

-53The evolution of the conical flow pattern due to a further increase of incidence depends on the influence of viscosity. For inviscid flow at high incidences the vortical singularity as already suggested by Ferri (1951) lifts off from the cone surface resulting in a saddle point at the leeward generator and a nodal point in the symmetry plane above the cone surface (Fig. 2.3d). However, if viscosity is taken into account its influence tends to obscure the lift-off phenomenon, as is observed in the experiments made by Bannink & Nebbeling (1978). These experiments show that at high incidences flow separation leads to the generation of a vortex system on the leeward side of the cone (Fig. 2.3e). The separation point may be considered as the viscous analogue of a conical stagnation point having a saddle type structure with oblique separatrices. Similarly, the center of the vortex may be seen as a conical stagnation point with a spiral structure. If the incidence is increased still further the spiral structure persists but the flow pattern above the vortex changes due to the appearance of a saddle and a node in the symmetry plane (Fig. 2.3f)- If we compare the 'viscous' and the 'inviscid' route in the development of conical flow patterns we observe that viscosity can have a significant influence; it causes separation, vortex formation and a delay of the lift-off phenomenon. Moreover, we observe in the 'viscous' as well as in the 'inviscid' scenario that the flow topology changes at certain combinations of the global parameters (incidence, Mach number, semi apex angle of the cone), suggesting that at the corresponding parameter values the conical flow pattern becomes structurally unstable and bifurcates.

Another example of changing flow topology may be found in the flow around an elliptical cone if the incidence is varied. A scenario of inviscid flow patterns for increasing incidence as obtained by numerical calculations (Grossman (1979)) is shown in Fig. 2.4. At zero incidence the flow pattern contains four conical stagnation points (two saddles and two nodes) on the body surface. The nodes N. and N- occur on the minor axis and the saddle points S1 and S_ are on the major axis of the ellips (Fig. 2.4a). Variation of incidence in the plane of the minor axis does not disturb the symmetry of the flow pattern with respect to this plane. For small incidences the saddle points S 1 and S_ move towards the compression side of the cone whereas the nodal points N. and N_ maintain their original positions on the minor axis (Fig. 2.4b) . A further increase of incidence causes the coalescence of the saddle points S- , S p and the node N.. After coalescence a topologically different flow pattern results with two conical stagnation points at the cone surface, the original node N_ at the leeward

-54-

meridian and a saddle point S a t the windward meridian (Fig. 2 . 4 c ) . At the p a r t i c u l a r incidence where t h i s transition occurs the conical flow pattern i s s t r u c t u r a l l y unstable and bifurcates.

increasing incidence

_^.

Fig. 2.4. Conical flow patterns for elliptic cone at incidence. On the basis of the previous discussion we summarize the following observations: - Conical flow patterns generated by conical bodies with relative simple cross sections may contain points where the cross flow velocity vanishes, the so called conical stagnation points. These points can occur on body contours as well as 'free' singularities in the flow field. - Various types of flow patterns may be observed in the vicinity of conical stagnation points. If viscosity effects are important there appears, next to the more familiar orthogonal saddles and nodes, also oblique saddle points and vortices (foci?). Furthermore, a conical stagnation line or a starlike node can be encountered in special cases. - Finally, there is a strong evidence that conical flow patterns may be structurally unstable leading to local bifurcation involving the generation or disappearance of conical stagnation points. Due to their sensitivity to perturbations the corresponding structural unstable flow patterns are hardly to realize in experiments or by numerical computations. These observations motivate our systematic investigation of flow patterns near conical stagnation points as will be performed in the following paragraphs.

-55-

2. Local conical stagnation point solutions in irrotational flow*)


2 1 l^_Conical potential_equation
Consider the flow of an inviscid, non heat conducting, perfect gas with ratio of specific heats Y = c /c . p v The flow is assumed to be irrotational so that a velocity potential $ may be introduced by V$ = V(u,v,w). The conical similarity allows the introduction of a conical potential F(n,t;) from which the velocity components may be derived by V = V(x F(n.O) (2.2)

From the conservation laws (mass and momentum) and the pressure-density relation for isentropic flow, it may be derived that F satisfies the nonlinear secondorder partial differential equation, Bulakh (1970) A F + 2B F _ + C Fr_ = 0 (2.3)

nu
with

nC

CC

A = c2 (1+n2) - (v-un)2
B = c2
2

nC - (V-UTI) (w-uO

C = c (1+t;2) - (w-uq)2 where the speed of sound c is related to the velocity by c2 = ^ 2 (q2 - u2 - v2 - w 2 ) max ' (2.4)

and q is the maximum speed, which we assume to be constant throughout the max ^ flow field. Equations (2.3) and (2.4) allow velocities to be nondimensionalized by q ; as a result we put q = 1 so that (2.4) becomes . max max
C

= - ^ (1 - u2 - v2 - w2) " 2

The velocity components u.v.w follow from (2.2) as u = F - n F Tl v = F w = F"

CF

*) A concise version of this paragraph was published by Bakker, Bannink & Reyn in the Journal of Fluid Mechanics (1981), Vol. 105, PP- 239-260.

-56Then the conical streamlines obey the equation


(1+T1 2 )

I> = V -

U71 =

-T) F

TlC

q = w - uq = -q F + nq Fn + (1+q*) F In the course of this study it appears also convenient to use polar coordinates n = p cos <p, C = P sin < p(0 < p < 2n, p > 0) in the cross flow plane X; then the velocity components become
u = F - pF , P v = F cos < p - - F s i n <p, p < P ' sin < p + F c o s <f P < P

w = F

and (2.3) may be written as


2 fcMl+P ) "L r [pF-(l+p*)F F v r * v r ' ' pJ y) ' pp

+l 2(F-(p +'-) p J F -2 V <PF <P 1F + ,r l If- F p p p<p p

(c*'-}(J + iF p ).0
The conical streamlines are the integral curves of the system p = (1+p*) Fp - pF 1 p2

(2.5)

q >

2.2. Conical_stagnation_goint solutions Approximate solutions of the conical potential equation (Eqs. 2.3. 2.5) near conical stagnation points are obtained by Bakker, Bannink & Reyn (1981). These solutions are locally valid and are found in the form of a series expansion in terms of the distance to the conical stagnation point. The coordinate system is chosen with its origin in a conical stagnation point. The following series expansion of the conical potential is used F = Fo(l + p n F ( < p ) + pm F ( < p ) + o(pm)}, 1 < n < m

-57where F i s a c o n s t a n t which equals the non-dimensionalized r a d i a l velocity o component in the conical stagnation point. The magnitude of F c has a s i g n i f i c a n t i n f l u e n c e on the c o n i c a l streamline FsF p a t t e r n . For supersonic conical stagnation points we have c < |F | < 1, where

i s the speed of sound a t the conical stagnation point. The solution

yields a uniform p a r a l l e l flow with a conical stagnation point at p = 0. The c o n d i t i o n n < m provides p F (q>) as the leading term in the expansion and the condition 1 < n assures that p = 0 i s a conical stagnation point. If Eq. (2.3) i s evaluated in p = 0 (or n = q = 0) the r e l a t i o n ^(0.0) FK(0.0) = 0 results which shows that near the conical stagnation point Eq. (2.3) is nearly the Laplace equation, which is also satisfied by the velocity potential $ in incompressible plane flow. Therefore, the proposed expansion for F is suggested by incompressible plane stagnation point solutions which are given by $ = a p cos (n<p + i | ) ), n > 1 n n where a , t | i are constants, they represent the well known flows in corners with opening angles w = n/n. If the expansion for F is substituted into the conical potential equation, Eq. (2.5) and the results are ordered with respect to powers in p, the coefficient of the lowest-order term equated to zero yields F" + n* F = 0 n n with t h e s o l u t i o n s F (<P) = e cos(n<p + < > | ), n > 1 n n n (2.6)

where e and Tt | > are arbitrary constants. n n Thus, this term exactly equals the stagnation point solution for plane flow. Since a freedom is left in the choice of the coordinate system, we may still rotate it around the x-axis such that t ( i =0. n

-58-

When t h e n e x t h i g h e r o r d e r
\y~- m = mc mc = n+2 s) mc=3 1 - 2
v>X

terms a r e

w r i t t e n o u t s e v e r a l c a s e s f o r n and m have t o be d i s t i n g u i s h e d . The domain i n the n,m-plane for which i t i s p o s s i b l e t o determine t h e f u n c t i o n s F (<p) and F (<p) nv ' mv

^ m=n

is shown in Fig. 2.5- After equating the coefficient of the next highest order term to zero we obtain the following result

\ /
i . i

F" + m2 F m m

0,

n <m< m m= m c f o r dif-

F" + m2 F = R (F , F ' ) , m m n n n Fig. 2.5. Possible values of m and n for which F v (9) and F v( < p ) n ' m exist. 1 < n < 2: n = 2: n > 2: where R and m a r e s p e c i f i e d , n c f e r e n t n , as follows:

2 R = n(n-l) M F (n2 F2 + F ' 2 ) n o n n n '

, m = 3n " 2 c
m = H

= - 2 F _ [ l - M 2 (1-2F_) 2 ] - 2M2 F ; 2 ( 1 - F _ ) , o 2

2 R = n(n-l) M F n o n

, m = n + 2
C

The q u a n t i t y M i s t h e Mach number i n t h e c o n i c a l s t a g n a t i o n p o i n t . I f m > m we o b t a i n R = 0 which y i e l d s , u s i n g ( 2 . 6 ) , F ( < p ) = 0 i m p l y i n g m = m i s an upper bound f o r m beyond which no s o l u t i o n s f o r F (<p) e x i s t . c m I f F (<p) e x i s t s we find t h e following s o l u t i o n s F m (*) = p m cosfm* X m ), F (<p) = p m m cos(m<p + X ) + T cos(n<p) + 6 , m m m n <m < m
(2.7)

that

= m

The c o n s t a n t s 6 and X may be chosen a r b i t r a r i l y , whereas t h e c o n s t a n t s T and m m m < 5 a r e r e l a t e d t o n , e and M2 as s p e c i f i e d i n t h e next t a b l e . m n o

-59-

range of n

m c

T m

6 m

1 < n < 2

3n - 2

n ' M2 e ' o n 4(2n-l) c 7 l(M2 - 1 + t e 2 M2 1 6 o n o ' n ( n - l ) (1P-1) E n

0
G2 M2

n = 2

n 2 0

n > 2

n + 2

2(n + 2)

Table: m , T f,< 5 . c m m The solutions, discussed so far, are obtained for integer as well as for noninteger values for n and/or m. For non-integer values of one of these exponents, the solutions are not 2n-periodic; implying that it is impossible to fill out a full neighbourhood of the stagnation point with solutions such that the velocity is continuous. This implies that solutions which are not 2TI-periodic have only physical relevance for conical stagnation points which are located on the contour of a conical body. Domains of multiplicity of the solutions are then masked by the interior of the body so that a single-valued flow field outside the body remains. In general this results in flow fields over partially smooth bodies with discontinuities in the slope where conical stagnation points occur. With the aid of the listed solutions for the conical velocity potential the conical streamline pattern near conical stagnation points may now be determined. In addition, the pressure distribution will be determined and compared with the streamline patterns. In isentropic flows the pressure p is solely a function of the speed |v| and follows directly from Bernoulli's equation

T-l p

= const.

and Poisson's relation

p = const, p

-60If the constants are evaluated in the conical stagnation point these equations may be written as T-l Y
=

,p_i l

1 -

(u*

pJ

1 - u2

* v8

w' )

(2.8)

where zero subscripts refer to conditions in the conical stagnation point. Equation (2.8) shows that the isobars in the cross flow plane 2 coincide with constant velocity curves so that they can be determined from ^ lyl2 -uuq - w g - ww^ uu n + w r\ + wwn and u , the isobar pattern is obtained

(]
v

dCp=const.

3 |V|" 3n

Using u = F - nv - qw to calculate u from

_ -(v-un) v c - (w-uC) w q d/p=const. (V-UTI) v + (w-ut;) w

(2.9)

From this equation it may be observed that the isobar pattern in the cross flow plane 2 has a singularity in any point (rut) if (v-un) v (v-un) v
' Tl

+ (w-uO w = 0 + (w-uO w = 0
Tl

thus either v w v^ w,

=0

or

= 0

A conical stagnation point always corresponds to a singularity in the isobar pattern. The reverse statement is not true, since if the second condition is satisfied (2.9) has a critical point but q is not equal to zero.

-61-

3.Classification of conical stagnation points in conical flow


3il1_First-order_conical_stagnation_points With the s o l u t i o n s , given i n paragraph 2, we w i l l i n v e s t i g a t e t h e t h r e e l < n < 2 , t h e d e s c r i p t i o n of t h e r e s u l t i n g c o n i c a l s t r e a m l i n e p a t t e r n s a n d o r i g i n a l p u b l i c a t i o n of Bakker, Bannink & Reyn (1981). Case 1 < n < 2: oblique s a d d l e p o i n t s S u b s t i t u t i o n of the s o l u t i o n s f o r F (<p) and F (<p) i n t o t h e e q u a t i o n f o r conical streamlines yields n+1 , , m+l> n c p cos n<p - p3 + 0(p ) 5 n . . , m. -n c p s i n n<p + 0{p ) the cases

n = 2 and n > 2 s e p a r a t e l y . The d i s c u s s i o n w i l l be c o n c e n t r a t e d on related

p r e s s u r e d i s t r i b u t i o n s ; for f u r t h e r d e t a i l s about the a n a l y s i s we r e f e r t o the

, 2 dq>

(2.10) v

'

I n t r o d u c i n g a parameter T along the s t r e a m l i n e s , we i n v e s t i g a t e t h i s e q u a t i o n as a system i n the (p,<p) p l a n e . Then dp -p- = n e d<p T~ = -n e


dT
n

3_n p cos nip - p

, m-n+1, + 0(p )
'

. , m-n. s i n n<p + 0(p )


kn

The s i n g u l a r p o i n t s of t h i s system on p s 0 a r e i n 9 = (k = 0 , 1 , 2, and t h e c o e f f i c i e n t m a t r i x of t h e l o c a l l y l i n e a r i z e d system has a

...) trace

p = i n e ( 1 - n ) , and a Jacobian q = - n ' e* . Since q < 0 the s i n g u l a r p o i n t s are h y p e r b o l i c s a d d l e p o i n t s and the h i g h e r o r d e r t e r m s do n o t c h a n g e t h e s u f f i c i e n t t o determine t h e approximate shape of the c o n i c a l s t r e a m l i n e s : p n s i n n<P = C, 1 < n < 2 saddle p o i n t c h a r a c t e r of t h e s e s i n g u l a r p o i n t s . Therefore t h e lowest o r d e r terms a r e

where C is constant along a streamline. This equation represents the streamline pattern for a flow in a corner with an including angle ( o = n/n; thus n/2 < < i > < n for 1 < n < 2. The flow in the corner has a saddle point behaviour and the point is called an oblique saddle point; its separatrices form the including angle < D = n/n at the conical stagnation point. Since 2n is not an even multiple of n/n an oblique saddle point can occur

-62o n l y on a b o d y . c o n t o u r w i t h one o r more s t r e a m l i n e s i n t h e s o l u t i o n c o i n c i d e with the body. The p r e s s u r e d i s t r i b u t i o n near the conical s t a g n a t i o n p o i n t may be obtained from ( 2 . 8 ) as T-l T
p

that

1 = 1

T-l , , 2n-2 n+m-2). n, r - M2 e2 n2 K p + 0(p ) v 2 o n '

The pressure attains a maximum in the conical stagnation points and the isobars are to a first approximation concentric circles around the origin. In Fig. 2.6a the streamline pattern in a corner 0 i < p i n/n is shown. Fig. 2.6b gives the radial pressure distribution for various values of n, near an oblique saddle point. It appears that the pressure gradient at the conical stagnation point is singular for corner angles 2JI/3 < < i > < n, corresponding to 1 < n < 3/2, finite for n = 3/2 and zero for 3/2 < n < 2.
P

/Pc

P=7n

t1
-S^vniiu/niiiriDi, ^=0
1/ / / i

/,<n<2 n = V,

1<n<3/,

(a) streamline pattern

(b) pressure distribution

Fig. 2 . 6 . Oblique saddle p o i n t (1 < n < 2 ) . Case n = 2: s a d d l e , nodes so ollu uttiio n s F (<t>) and F (<p) f o r n = 2 i n t o t h e e q u a t i o n S u b s t i t u t i o n of the s the conical streamlines yields: - ( 2 e 2 + l ) C + (mF B s i n < > t + F ^ c o s <p) p m ~ + e 2 q ( n 2 - ( ; 2 ) + 0 ( p m + ) ( 2 e _ - l ) n + (mF cos < p - F' s i n 9) p"1 + e , n ( n 2 - q 2 ) + 0 ( p m + 1 ) ' d. m m d. (2.11) for

d dn

with 2 < m 4 . The e i g e n v a l u e s of the c o e f f i c i e n t m a t r i x of t h e l i n e a r p a r t : A yield 1,2 = - 1


2e

= A

+ A

~2'

= A

1A2 =

" '*e2

-63If (2.11) is written in polar coordinates and considered in the (p,<P) plane, application of the theorem of Hartman-Grobmann (Chapter I) yields that the linearized system has the same streamline pattern as the nonlinear system if |e|

* YIn the i\,Q plane the linear system reads

la
dt

( 1 + 2e2)T>,

dx

( 1

2e2)Q

which defines a continuous streamline pattern also in the n,Q plane. For |e_| = J - a degenerate singular point occurs which will be investigated in section 3-3 where higher-order conical stagnation points are treated. Equation (2.7), shows that for non integer values of m the function F ( < p ) is not periodic with period 2n; therefore, in that case it is not possible to cover a full neighbourhood of the stagnation point such that the velocity is continuous. If m = 3 or 4 and if all higher-order terms are 2n-periodic the conical stagnation point can be realized in the flow field away from the body. From the expressions: p = -2 and q = 1 - ^e* we observe that conical stagnation points are represented in the p-q diagram, on that part of the line p = -2 where pa S kq. It implies that in potential flow, conical stagnation points will be with nodal points if |e | < J - (q > 0) and saddle points is if |e 2 | > J - (q < 0 ) . In the special case e_ = 0 (q = 1) the nodal point is a starlike node streamlines approaching the singularity in different directions. Sketches of these first-order conical flow singularities are given in Fig. 2.7a; they may be observed in many practical flow situations such as the flow around cones, delta wings and axial corners.

( b )
L2<"/2

': ^B
e2=0

0<2<V2

t2>V2

Fig. 2.7. Structurally stable conical flow patterns, n = 2, |e_| = J . (a) conical streamlines, (b) isobars.

-64With Eq. (2.8) the pressure distribution becomes: T-l T .(E-) = 1 - 2(T-1) M Q c 2 [ ( e 2 - J-) n' ( e 2 J-) q ] 0 ( p m ) (2.12)

If |e_| > J-, and thus when a saddle point singularity of the streamlines occurs, the p r e s s u r e a t t a i n s a maximum in the conical stagnation point and the isobars are to a first approximation concentric ellipses around the origin. If 0 < | e 2 l * i " (nodal s i n g u l a r i t y ) t h e isobars are to a first approximation concentric hyperbolae with asymptotes given by
L
E 2

In the region within the acute angle between the asymptotes the pressure is higher than in the conical stagnation point, whereas in the other regions the pressure is lower; see Fig. 2.7b. For e_ = 0 the pressure does not change to the order mT/(T-l) of the distance from the conical stagnation points; this case actually corresponds to n > 2 which will be discussed hereafter. Case n > 2: starlike nodes Substitution of the solutions F ( < p ) and F (q>), as obtained for n > 2 into the equation for the conical streamlines yields , n+1 ., m+1. , -p' + n c p cos n q > + 0(p ) n S = d<p n . _, m+l> -n e p sin n<? + 0(p ) n which, in fact, is equal to Eq. (2.10) for the case 1 < n < 2 . Comparison of both equations however shows the d e c r e a s i n g i n f l u e n c e o f the c term with respect to the radial velocity term (-p*) when n > 2 . Introducing the parameter T along the s t r e a m l i n e s w e i n v e s t i g a t e (2.13) as a system in the (<P,a) plane where a = p Then m-n 9 - -n c sin n<p + 0(a n ) 1 m-n ) (a 0 ) .

(2.13)

1 + ^2

a - -(n-2) + n(n-2) e a cos nq> + 0(a

-65The line o = O contains only regular points, resulting in a single integral curve through any point of this line. Correspondingly, the streamlines approach the origin in the cross flow plane S from all directions. To the order considered, the lines < p = k n/n (k = 0, 1, 2, ..) are P=Po
P = Po

conical streamlines dividing the flow field into sectors with an opening angle t o = n/n ( t o <n /2). The conical streamlines in each sector are curved at the singular point and the curvature is opposite in adjacent
mum

sectors. The flow in two adjacentsectors is shown in Fig. 2.8. Similar to the previous cases, this type of conical stagnation

Fig. 2.8. Starlike node at n > 2.

point in general cannot occur as a 'free' singularity in the flow field, certainly not if n is not an integer. The pressure distribution, which is given by the relation T-l (2-) p o = 1 + (Y-l) M'(n-l) e p n cos n q > + 0(pm) + 0(p2n"2) o n

shows a saddle behaviour near the conical stagnation point (see Fig. 2.8), with sectors correlated with the sectors in the streamline pattern. A special example of this singularity is the conical stagnation point in a uniform flow. Then E = 0 and the isobars shown in Fig. 2.8 do not apply since in this particular case the pressure is constant throughout the flow field. The previous investigation of first-order conical stagnation point singularities shows us that in irrotational conical flow different types of singularities may be encountered such as oblique saddle points, saddles whose separatrices are perpendicular to each other, nodal points and starlike nodes in corners with an opening angle less than n/2 rad. However, other types that probably could be expected such as spiral points and oblique saddles in acute corners, as mentioned in paragraph 2.1, are not found in this classification. In this respect it must be emphasized that the results are obtained for irrotational flows; the extension of the classification for nonisentropic conical flows is subject of current research.

-663;2i_Irrotational_attachments_and_segarations It is of interest to discuss attachment and separation phenomenae in inviscid flows along a smooth conical body from a point of view of first-order singularities. Such situations are likely to occur in many flow problems, especially in flows around bodies at high incidences. Let us consider the details of the flow near a conical stagnation point which lies on the contour of a smooth conical body in such a flow. The smoothness of the contour implies that only conical stagnation points can occur which belong to the class n = 2. Let the body contour in the cross flow plane I be defined by t ;= t ; (TI) and assume that the conical stagnation point is located on i ; at n = C = 0. The body contour near the origin may be approximated by t ; = 0(TI J ). Referring to the first-order singularities, obtained for n = 2, the following distinct streamline patterns may be considered. e_ > J a saddle point of attachment. the body contour (tangent node). G_ = 0 J - < e. < 0 c- < J a starlike node. a nodal point having an infinite number of streamlines perpendicular to the body contour (normal node). saddle point of separation. 0 < e < J - . a nodal point with an infinite number of streamlines tangent to

Let us first give attention to the pressure distribution over the body surface in relation with the cross flow streamline pattern. From Eq. (2.12), we obtain for the pressure over the surface: T-l T (jj-) = 1 - 2(T-1) M^ c2(e2-i) p o

n2 0(n)

This equation shows that the pressure has a local minimum if a tangent node occurs, whereas for a normal node and a saddle point (attachment or separation) the pressure attains a local maximum. In the special case of a starlike node the pressure remains constant to second order. Obviously only tangent nodes can be determined uniquely from a surface pressure distribution. To describe all possible streamline patterns in a unique sense the pressure behaviour perpendicular to the surface has to be known as well. A concise result may be obtained if the second order pressure derivatives are evaluated in the conical stagnation point. From Eq. (2.12) we obtain:

-67-

*ni, = 0

(0 0)

' " S * 1 " 2c2>

316 (0,0) = -Hz (l 2c-) 3C


M2 in

where is the pressure nondimensionalized by the dynamic pressure J -T p the conical stagnation point, p = .
y

,TpoM' o

If e_ is eliminated the following relation between p (6 - 6 ) 2 + 4(6 + p )= 0

and 6

results

Fig. 2.9. Second-order pressure gradients in conical stagnation point (n = 2). From Fig. 2.9, which shows this relationship, it is easy to conclude that a local maximum of the surface pressure {p a normal node if * > 0 < 0) corresponds to:

- a saddle point of attachment i f T " > 1


Tin

- a saddle point of separation if r " < 1


Jin

Furthermore it may be noticed that p nn isentropic conical flows.

and 6^,. never exceed the value +\ in CQ

-68The properties of conical stagnation points on a smooth body may also be studied in relation to the spatial inviscid streamline pattern on the body surface near those points. From the solution given by Eq. (2.6) the velocity components may bederived as u = u Q (1 - e2(n - < ; ' ) } 0(p) v = 2u w = -2u c 2 n + 0(p2)
G2

q + 0(p*)

Since the body surface i s approximately coincident with the plane q = 0, the s p a t i a l surface streamlines follow from

dx'q=0

1 + 0(n 2 )

which may be integrated for small n to obtain the spatial streamline pattern close to n = 0 where zero cross flow appears. The correspondence of surface streamlines with cross flow streamline patterns is shown in Fig. 2.10.

Fig. 2.10. Surface streamlines and corresponding cross flow pattern. It appears for a tangent node that the surface streamlines, diverge from the attachment line (x-axis) forming a pattern with a so called 'gothic' behaviour. In the case of a starlike node the surface streamlines are to a first approximation parallel to each other.

-69The normal node and the saddle point of separation generate spatial surface streamlines which converge to the ray with zero cross flow (x-axis). In the nodal case the rate of convergence is less than in the saddle case. A saddle point of attachment is associated by a surface flow pattern diverging from the attachment line.

Finally it is of practical interest to obtain the correspondence of the cross flow streamline pattern with the spatial surface flow direction near the conical stagnation point. The surface flow direction, denoted by * , is defined as the angle of the velocity vector with the local ray emanating from the conical center, see Fig. 2.11, then there is obtained tan *

u + vn

The angle * passes through zero i n a c o n i c a l s t a g n a t i o n p o i n t . The r a t e a t which t > | changes along the surface i s determined by the p a r t i c u l a r cross flow s d* pattern since -z = 2e_ - 1, so the cross flow pattern a t a c o n i c a l s t a g n a t i o n n d < > | p o i n t may be uniquely c h a r a c t e r i z e d by -z ( s e e F i g . 2.11) as follows
conical center

local ray

-2

"1

*s

W/V/V.

\\
Fig. 2.11. Directional gradient of surface streamlines.

dV > 0: saddle point of attachment


di|i

-1 < -; < 0: tangent node d* s dn


dt|>

-1: starlike node

~2 < -z < -1: normal node dn d^ dn < -2: saddle point of separation

-703^3i_Higher-order conical_stagnation_points Referring to section (3-1) of this paragraph it may be concluded that higherorder conical stagnation points occur for n = 2 and |e.,| = . If |e_| = J - one of the eigenvalues of the locally linearized system is equal to zero and the nonlinear terms in Eq. (2.11) cannot be neglected when considering the streamline pattern. The character of these higher-order conical stagnation point singularities will now be investigated under the restriction that only 2n-periodic solutions of the conical potential are considered. Thus only those higher-order singularities are discussed that can occur on smooth body contours or as 'free' singularities in the flow field. Then, the following solution for the conical potential F is obtained from Eqs. (2.6) and (2.7)
F = FQ(1 + E2(TI2 - q2) Pl (n 3 - 3TK Z ) - P2(3nJq - t;3) +

(2.14a) -2o(n'i; - Tiq') + ^n* - yi ? where e_, p . , f> , 6 depend on < 5 _ by


1 1

+ 6?V + 0(p ))

and 62 a r e arbitrary constants; the parameters 6. and <5,

1 = I 62 + F t M o ( 1 " 22>* - ! )
(2.14b)

3 = I 62 - T K{1

+ 2E )2

' !)

and M i s the local Mach number a t the conical stagnation p o i n t , i t depends on F as o

M b = c 4 - T-l o
2

F2

F2

1 - F2 o

The c o n i c a l s t r e a m l i n e p a t t e r n near conical stagnation points consists of the phase t r a j e c t o r i e s of the system that follows i f Eq. (2.14) i s s u b s t i t u t e d the conical streamline equation n = v - un, < ; = w - uQ: into

-71+ u ^ Q g 2 (?) v + 1 Ti3_i e 1 + 0(p*) (2.15) q = w z = -(2e2*l)C " u x nC + /


( 2-i:2) +

n = v z = (2e2-l)n +

(TI-C*)

"

g z (?) v + i S 3 "* ^

(P*>

The coefficients u . , v. and v* are new unknowns instead of p. and 6.. Their dependence on p. and 6. is irrelevant for the analysis. Of more importance are the mutual relations between v. and v* which follow from Eq. (2.14). v = -v = -v^ = v 2
V

3=

4e

2M o ( 1 ~ ' 2 e 2 ) ' (2.16)

M* +

v* = - 4 E 2 M^(l + 2 E 2 ) 2

v* -

v 1 = -8e 2 M^(l + 4 E 2 ) 2

-4E2

The flow patterns at E p = +J- may be derived from those at E_ = -J- by performing the transformation
T>

* -<; n

v > v*
v

l + vl
V

^1 * ^2 p 2 * -n 1

34

v + v 3

which shows t h a t the nature of the degenerate singularities a t ., = +J- and at E- = -J- i s similar and t h a t a r o t a t i o n of it/2 rad transforms them i n t o each o t h e r . Therefore, we r e s t r i c t the discussion on higher-order s i n g u l a r i t i e s to one value of e_ only, say e_ = -J-. For e_ = -J-, system (2.15) has a degenerate singularity at (0,0) with one zero eigenvalue. Then Andronov's Theorem 65 about t h i s type of m u l t i p l e s i n g u l a r p o i n t s may be applied (c.f. Chapter I) from which follows that n = 0 on a curve which i s approximated by l n = f ( c ) = - i p :* + o(;)
U

-72-

and that on t h i s curve i = Am qm 0(q m+1 ) = - / < : (g 1 5 1 ) C' 0 ( O ) (2.17)

According to Andronov different degenerate s i n g u l a r i t i e s can be d i s t i n g u i s h e d

Case: u 2 * 0, saddle-node For u_ * 0 t h e r e follows m = 2 and 4 = -u_/2 so that the multiple singular point i s a saddle-node, i t has one parabolic sector (nodal type) and two hyperbolic sectors (saddle type), see Fig. 2.12a. An unlimited number of streamlines, constituting the nodal p a r t , approach the c o n i c a l s t a g n a t i o n p o i n t along the direction n = 0 whereas two single streaml i n e s approach the saddle-node along the d i r e c t i o n t; = 0; they are convex towards the nodal part of the saddle-node. This implies that if a saddle-node i s f i t t e d on a body surface, the nodal part occurs for convex surfaces whereas for concave surfaces only the saddle part appears. The particular streamline which separates the parabolic s e c t o r from the hyperbolic sectors has the approximate form
q =

" g u 2 n*

0(n,)

The o c c u r r e n c e of a saddle-node in cross flow s t r e a m l i n e p a t t e r n s may be expressed in terms of the cross flow velocity derivatives as follows.
3w = -J-(l + - ) 3v = J-(l + - ) , 32w u _ = - jpr~

From Eq.

( 2 . 1 5 ) we o b t a i n e

which

implies that a saddle-node occurs if the conditions: 1. v 2 = wz = 0,

aJw

are satisfied in a common point in the flow field.

-73-

(a)

(b)

Case_p_ = 0; t o p o l o g i c a l s a d d l e s , topological nodes For u 2 = 0 t h e r e follows from Eq. (2.17) m = 3 and Am = (^- + g) or, expressed in cross flow derivatives, (c.f. Eq. (2.15))
a z w,
A

P=Po P=Po

H Un 3 ^

6 3C'

saddle-node

YL
~\

typological saddle

topological node

From Andronov's theorem (Chapter I) i t may be concluded t h a t the s i n g u l a r point i s a topological saddle point if A >0 and a topological node i f A <0; sketches of them are shown i n F i g . 2.12a. I t i s possible to extend the c l a s s i f i cation to higher-order s i n g u l a r i t i e s by investigating the p a r t i c u l a r case A _ = 0. However the a c t u a l occurm=3 rence of such s i n g u l a r i t i e s i n r e a l flows becomes l e s s l i k e l y s i n c e an increasing number of r e l a t i o n s between higher-order derivatives of the cross flow v e l o c i t y components have to be obeyed by the flow. Therefore the case 0 i s left undiscussed. 'm=3 The i s o b a r p a t t e r n near the higherorder s i n g u l a r i t i e s (e_ = -J-), d i s c u s s e d so f a r , may be obtained by s u b s t i t u t i n g Eq. ( 2 . 1 4 ) i n t o Eq. ( 2 . 9 ) . The i s o b a r s a r e then the integral curves of the following system

Fig. 2.12. Typical structurally unstable conical flow patterns. (a) conical streamlines, (b) isobars.

-7f-

= i - ^ Pj i

- \ u2

nt: +

'

+ 2 B

T,3

~i

ql

where T is a parameter along an isobar, A

and B

are coefficients which depend

on u , v and v*. They will be specified in the course of the analysis if their respective values are needed. Both eigenvalues of the linearized system near (0,0) are equal to zero so that Androhov's theorem about this type of multiple singular points may be applied (c.f. Chapter I). The curve on which -j* = 0 is
u l i t = - ,j Q' + 0(q)

and on this curve

g
where

= Ak qk 0(qk+1) .%?

(Ji-A3) q' 0<O)

12

Hence it follows that for u_ * 0, k = 2 which implies that the isobars form a cusp near the singularity. As we have shown before, the case u ? * 0 corresponds to a saddle-node of conical streamlines: the isobar pattern is sketched in Fig. 2.12b. For u_ = 0, there is obtained
u l 1 k = 3 and A k = g- - A 3 = - ^ A m

implying that the isobar pattern exhibits a center for A

< 0 and a topological

saddle for A. > 0. k Since A, and A have opposite sign we obtain as a final result that a topologiK m cal node of conical streamlines (A < 0) corresponds to a topological saddle of isobars (A. > 0) and a topological saddle point of conical streamlines (A > 0) gives a center point of isobars (A, < 0 ) ; sketches of these pressure figures are shown in Fig. 2.12b. In these pictures, regions with a pressure exceeding that in the stagnation point are indicated by a plus sign, whereas a lower pressure is denoted by a minus sign.

-754. Analytical unfoldings in conical flows

The second-order system Eq. (2.15). which describes the behaviour of conical streamlines near conical stagnation points, contains several unknown parameters (e_, u-, v., v*) which define a multiple parameter family of functions in the right-hand side of the system. These functions are represented by Taylor series wherein these parameters may be identified as spatial derivatives of the cross flow velocity components in the conical stagnation point. The topological structure near a conical stagnation point will depend on the actual parameter values of c_, u., v., \ > * which can also act as bifurcation parameters. Since higher-order stagnation points occur at |e_| = J - (which are saddle-nodes if Up * 0 and topological nodes or topological saddles if j i _ = 0) , the bifurcation parameters are obviously e_ and y_ with bifurcation values |e ? | = \ and Up = 0, respectively. The right-hand side of Eq. (2.15) is represented by a Taylor series, therefore we consider only bifurcations caused by analytic perturbations.

The bifurcation of a saddle-node is described by at least one parameter (e_) whereas the bifurcation of a topological node and a topological saddle point is governed by at least two parameters (c? and u~) . So far the parameters in Eq. (2.15) are recognized which may cause bifurcation. However, the bifurcation processes that can be treated now are restricted to flow patterns which contain after bifurcation at least one conical stagnation point. This property of the bifurcation process is due by the fact that Taylor expansions are performed with respect to a conical stagnation point so that the system has always a singular point at the origin (0,0). This restriction is rather unfortunate, because it prevents the description of spontaneous generation of conical stagnation points; furthermore it prevents also the observation of flow phenomenae where two or more singularities merge together and disappear after coalescence. This difficulty may be circumvented if the bifurcation of degenerate singularities allows for nonzero cross flow at the original location of the higher-order singularity. This means that the magnitude of the cross flow velocity |q | at n = Q = 0 has to be taken as an additional independent bifurcation parameter.

-76Together with c_ and u ? we expect a codimension-two bifurcation for the saddlenode and a codimension-three bifurcation for the topological node and the topological saddle. For a proper description of these bifurcations the system of Eq. (2.15) is not appropriate since the third bifurcation parameter |q | does not appear in the right hand side of (2.15). A modified formulation follows if in Eq. (2.1) conical flow solutions are used which are obtained by expanding F(n,0 near a regular point in the flow field. In general this expansion may be performed near the origin n = C = 0 so that F(n,0 may be approximated by: N 2 p n Fn(<p) 0(pN+ )) n=l

F = Fjl +

(2.18)

Among the unknown functions F (<p), (n = 1, .. N) (which will be solved up till N = 4 in the next paragraph), only F. ( q > ) contributes to the cross flow velocity q_ at the origin, since: dF *

lgzM(o.o) = P{F (^)

(2.i9)

4;21_Approximate_solutions_near_regular points To construct approximate solutions of the conical potential equation, Eq. (2.5) near points with nonvanishing cross flow and to be in accordance with stagnation point solution developed in section 2.2, we will expand the conical potential in terms of the distance to the origin TI = q = 0 by the power series: N ^ I p n Fn(<p) + 0(pN+1)) n=l

F = F (1

(2.18)

If t h i s equation i s s u b s t i t u t e d i n t o Eq. (2.5) and the result i s ordered in powers of p, ordinary d i f f e r e n t i a l equations for the functions F (<p) a r e obtained. The equation for F. (<p) gives
Y-l

(1 " P - F ^

, , ,

^ F i ) } ( F ; ^ } =0

T+l

-77which will be satisfied by two different types of solutions F^Ut) and


M - p*
v

= a 1 cos(<p - i^)

(2.20)

Fj2)(<P) = - ^ s i n o
(2)

( \ M - <P0)

(2.21)

The second solution F. (<t) is not 2n-periodic and should therefore be omitted here. The solution describes the conical analogue of a Prandtl-Meyer fan and represents a possible flow near a discontinuity in the slope of a conical body. The first solution F^ (<P) contains two arbitrary constants a- and i|)., which may identified with the magnitude and direction of the cross flow velocity at the origin, respectively. The relation of |q | with a1 is given by: |gsl(0.0) = F Q VF2! + F'> = Fo.fll
The q u a n t i t y a . , which i s the b i f u r c a t i o n p a r a m e t e r we a r e l o o k i n g f o r , a l s o appear i n the s o l u t i o n s for F ? (<p), F_(<p) and Fj, (<(>), cf. Bakker (1984). Since a. has the b i f u r c a t i o n value a. = 0, only s o l u t i o n s for s m a l l a, appear t o be s i g n i f i c a n t . The l a b o r i o u s c a l c u l a t i o n s i n o b t a i n i n g t h e s e s o l u t i o n s , h a v e been made by Bakker (1984), h e r e we s u f f i c e t o l i s t the r e s u l t s F_(<p) = CJK cos *_ + cos(2<p - 211). + <|>2)}

(2.22)
will

Isil ,: where K is a function of the cross flow Mach number at the origin M lrt n c = c(U,U)
a' F2 MJ 1 0 c 2 ~ v (Y-l)(l-F ) - T a' F2 " 2 - M2 'l o 1 o c and e_ and t | > _ are arbitrary constants. We may use the freedom, still existing in the choice of the coordinate system to rotate it around the x-axis in such a way that t | > _-2 t | > , . = 0 resulting into (2.23)

-78F2(<p) = C 2 ( K cos 2 ^ + cos 2<p) Then the solution for F_(q>) and F^(<p) are found to be
a

(2.24)

Fnf'P) = c 3 1 sin 3<P + c

cos 3<P +

(T_1) ( 1 _ F f.

l c2 Fo

{(2E 2 -1) C O S <P C O S ^ + (2.25)

+ (2e2+l) sin <t sin * } + O(a^) and


F

1)W

= Cj^

s l n

***

42

cos

4<p - M' |

e2

+ |

e2{(l

+ 4c2)

M^, - 1} c o s

2<P + 0 ( a i )

(2.26)

where c,-, c__, Cj.. and Cj,_ are arbitrary constants. Note that the functions F_(q>) and Fj,(<p) are specified to the order 0(a*) and 0(8-) respectively; the specification of higher-order terms has been omitted here because their influence on the bifurcation process appears to be absent. The system defining the conical streamline pattern may now be written as n = P(n,0 = F Q {a 1 cos ^ + (2e2-l) n + j ^ j - + 2 nq + p 3 |- + 3 (2.27) q = Q(n.q) = FQ[&1 sin *1 - (2c2+l) q + 2 j " 3 * | I (?) v +1 q 3 _ i n 1 0(p) where 2 = e2(l + J - a* M^ cos 2 ^ ) + O(a^) e 2 = e2(l - J - a^ M^ cos 2*1) + 0(a'1) ^1
= w +

3 nq + ,, - +

+ 3a

l E2(2c2 "

1) M +

o
M

COS

*1

+0(a

i) ^al^

P 2 = u 2 + a 1 e 2 ^ 2e 2

Sin

*1

jL = -Pj^ + a c 2 (2e 2 - 1) M^ cos ^ + Ofa^) u/j = -u 2 + 3a x


e 2

(2E2

*'

M 0

sin

*i

< a i)

-79The c o e f f i c i e n t s v. and v* are related to v. and v* as v. = v. + 0(a.,) and v* =


1 1 1 X 1 1
v

1'

v* + 0(a.) respectively. Note that Eq. (2.27) reduces to Eq. (2.15) when taking a- = 0; representing the case that the origin becomes a conical stagnation point. * * 3i_ddle2node_bifurcation A saddle-node occurs in a conical flow pattern if the cross flow velocity components v and w_ satisfy the following conditions: 1. v 2 aw 2. ^
32WZ

= w2 = 0

{&1 = 0)

=0

(2 = - )

3- W T7T- * '0 "

(p, ^2

0)

The conical streamline pattern near this degenerate singularity may be obtained from (2.27) yielding * u 2 it ~ V 2
+

T I

= -2n + \xx Y

*p' ^ (2.28)

^= + u 2 2~~ ~ u l nC ~ y 2 I" + ( p ' )


where the immaterial (positive) constant F has been used to scale the parameter along the conical streamline. Bifurcation of the saddle-node occurs if the parameters e_ and a, are perturbed with respect to their respective bifurcation values c_ = J - and a, = 0. Assume that these perturbations are given by c and a which are small compared to unity, e_ = J - + e, a. = 0 + a. The unfolding of Eq. (2.28) which s a t i s f i e s the c o n i c a l flow e q u a t i o n s may be derived from Eq. (2.27) by substituting e ? = - + e and a1 = a. The qualitative behaviour of the flow p a t t e r n occuring a f t e r b i f u r c a t i o n mainly determined by the neighbouring conical stagnation points P . ( n . , i ; . ) might appear after b i f u r c a t i o n . To o b t a i n the l o c a t i o n s of t h e s e p o i n t s small a and c we use center manifold theory (Chapter I ) . is that for

-80-

For that, system (2.27) is extended with the equation a = 0 in order to obtain a system (in R 3 ) that contains c as the only parameter; a is now considered as an

additional independent variable.

For e = 0 this extended system becomes:


TIZ

q2

q = a s i n i\>1 a = 0

* |i

l ^ nq - u 2 f

* (3)

Tl2

C2

n = a cos j - 2n + u x 2~ + n 2 nq - p j | - + 0(3) which has a d e g e n e r a t e s i n g u l a r i t y o r d e r terms composed of q, a and n. The i n f l u e n c e of e i s of: analyzed in t h e c e n t e r manifold t h a t i s tangent t o the at ( 0 , 0 , 0 ) ; 0(3) denotes t h i r d and h i g h e r

c e n t e r eigenspace /0 0 0 sin 0 cos $.

*1

o]
with eigenvalues A. = A_ = 0, A- = -2 -2

The corresponding eigenspaces are

E c = (0,2,cosily) X(l.O.O),

E s = (0,0,1),

Eu = *

c c The center manifold W : n = h(q,ct), tangent to E satisfies the equation /3h 3q 0 3h 3a 0 /.\ n

i;

with the boundary conditions: h(0,0) = h (0,0) = 0, h (0,0) = J - cos i^.

The approximation of the center manifold near (0,0,0) becomes

W : h(q,a) = h^

+ h 2 a 2 + h-aq + h^q2 + 0(3)

where 0(3) denotes terms of 0(q', q 2 a, qa 2 , a 1 ) . The coefficients h , h-, h and lu are determined as

V cos *.

-81v U l 2 l 2 hp = -T7 cos il), - K~- sin ij). cos t l > . - g - sin 2 1 ( 1 . U

U1 sin Tp^ + y 2 cos * 1

S ="T
The essential features of the unfolding follow i f (2.27) i s projected on W :
pa

C = a sin *. + 8

or cos 4

(*j) - (2e - 2^" cos * 1 ) C " P 2 f

+ 0(3)

(2.29)

which shows that for a = 0 and e = 0 the C2-term i s the lowest-order nonvanishing term implying a codimension-one bifurcation so t h a t only one p e r t u r b a t i o n parameter should be necessary to decribe the bifurcation in a persistent way. However, Eq. (2.29) a l s o shows t h a t the p h y s i c a l i n t e r p r e t a t i o n of t h i s b i f u r c a t i o n needs the use of two parameters (a and e ) . The locations of the neighbouring conical stagnation points will depend on these parameters and may c c be obtained from the expression for W and the projected system on W . For small values of a and c the approximate l o c a t i o n i s given by the leading terms: -2c 2 Je 2 + J- u- a sin tK
C

i f sin iK * 0, i = 1,2

(2.30a)

n. = i- a cos
and

h -4

^IV

C = -CU:

Uja) J(4e + uj a) 2 + (u 2 a) 2
2P,

i f sin t|). = 0 (cos 4^ = +1), i = l , 2 (2.30b)

ni =

L a

If sin I|I * 0, the saddle-node f a l l s apart into two conical stagnation points if 2 the q u a n t i t y u_ a sin ty-, which we w i l l call a, exceeds the value -2 e . The type of these points follows by evaluating the t r a c e p and the Jacobian q at these p o i n t s ; i t appears t h a t the plus sign in Eq. (2.30a) corresponds to a

-82stable node and the minus sign to a saddle point. For a = -2 e2 a nongeneric bifurcation occurs; the saddle-node remains present and is shifted with respect to its original position. If a < -2 c2 generating any neighbouring singularity. A qualitative impression of saddle-node bifurcation in conical flow is given in Fig. 2.13 which shows the bifurcation set together with conical flow patterns. If sin *. = 0 saddle-node bifurcation will occur under the forced condition that the cross flow velocity component w stable node. remains zero. As a consequence, always two conical stagnation points will appear after bifurcation: a saddle-point and a the saddle-node disappears without

'a=-2e2

Fig. 2.13. Bifurcation of the saddle-node. Let us now discuss further the case sin i | ) . * 0 and see how saddle-node bifurcation may be used to explain some transition phenomena in conical flows. The bifurcation set, Fig. 2.13, shows that a saddle-node may bifurcate into two different ways. The first possibility is its falling apart into two structurally stable singularities and the other is its disappearance leaving no singularities at all. The existence of these two bifurcation modes is the reason for two different transition processes in conical flow patterns. Each of them may be identified with a traverse along a particular curve in the a,E-diagram. The first transition phenomenon occurs if one moves for example along a = 0 from e < 0 to c > 0. The transition is marked by the fact that conical stagnation points

-83-

remain present during the t r a n s i t i o n . If e i n c r e a s e s from E < 0 to c > 0, a saddle and a node move together and coalesce into the saddle-node (e = 0 ) ; after coalescence the saddle and the node appear again and move away from each other. This type of bifurcation, which will be called: 'saddle-node bifurcation of the f i r s t t y p e ' , e x p l a i n s the ' l i f t - o f f ' phenomenon which was observed a t the leeward side of a cone at sufficiently high incidence {this c h a p t e r , paragraph 1.4). At l i f t off, the conical s t a g n a t i o n p o i n t a t the body surface in the leeward singularity plane would be a saddle-node; the nodal p a r t i s formed by the s t r e a m l i n e s outside the body surface and the saddle part i s found if the flow i s extended inside the body (Bakker & Bannink 197^) - A decrease of i n cidence leaves a nodal point on the cone surface and creates a saddle point in the solution inside the cone. If the incidence i s increased beyond the value where l i f t - o f f occurs the saddle-node f a l l s apart into a saddle point attached to the cone surface and a node moving away from the body. A d i f f e r e n t type of t r a n s i t i o n exists if we move in the bifurcation set (Fig. 2.13) from a < 0 to a > 0 along e = 0. We observe t h a t a r e g u l a r flow f i e l d e x i s t s as long as a < 0 and that only at a = 0 a saddle-node appears. A further increase to a > 0 resuls into the formation of a saddle point and a stable node. In conclusion, t h i s t r a n s i t i o n proces i s c h a r a c t e r i z e d by t h e simultaneous appearance of two conical stagnation points the ' b i r t h ' of which i s caused by the sudden occurrence of a degenerate saddle-node. This example of saddle-node bifurcation will be called: 'saddle-node b i f u r c a t i o n of the second t y p e ' . I t explains the observations of Bannink & Nebbeling (1978) that a saddle and a node suddely appear in the flow around a c i r c u l a r cone beyond a p a r t i c u l a r high incidence. 4 1 |l i _Bifurcation of topological saddle_goint A t o p o l o g i c a l saddle point occurs in a conical flow pattern i f the cross flow components v and w obey the following conditions
1. 2.

v 2 = wr = 0 (a t = 0)
3w

T
3 w
l

(e

2 -*>
l

34,

8C,

= 0 (u 2 = 0)
3JW 3'w

-84The conical streamline pattern near this degenerate singularity may be obtained from Eq. (2.27) as:
p

T I

= v s = -2n + Y

("'

" S')

+ 0(p

') (2.3D

Q = w_ = -p. nC + v 5- + 0(p' 1 6 where the immaterial (positive) constant F is included in the parametrization.

On the right-hand side, the term v* g- is written explicitly whereas the remaining third- and higher-order terms are denoted as 0(p') since their contribution to the qualitative flow picture appears to be a higher order effect. The detailed structure of the topological saddle point depends on the actual values of p. and v*. The topological saddle point has four separatrices, two of them approach to the conical stagnation point along the n-axis and the others along the ^-direction. The latter are curved and can be approximated by the parabola n = - r p1 t ; * . Bifurcation of the topological saddle point occurs if the bifurcation parameters c-, a, and p_ are perturbed with respect to their respective bifurcation values E- = -J-, a. = 0, v i p = 0. Analogous to the saddle-node case we assume these perturbations to be given by c, a and u which are small compared to unity; e2 = J -+
E,

a. = 0 + a, Up = 0 + u.

The unfolding of Eq. (2.31) satisfying the conical flow equations may be derived from Eq. (2.27) by substituting the perturbation parameters E, a and p. To obtain the locations of neighbouring conical stagnation points P.(n.,c;.) as a function of the perturbation parameters: a, E and p we use center manifold theory. For that system (2.27) is extended with the equation a = 0 in order to obtain a system (in R') that contains E and p as the parameters, a is now considered as an additional independent variable. For E = p = 0 the extended system reads: C = a sin t^ - p 1 nC + 0(3)

n2 n = a cos 4^ - 2n + p ^ ^ ^

C* p 1 |- + 0(3)

-85The influence of E and p will be analyzed in the center manifold that is tangent to the center eigenspace of 0 0 0 sin $1 0 cos *. 0 0 -2

The eigenspaces of the linear system are E c = (0,2,cos * ) X(1,0,0), E s = (0,0,1), E u = The center manifold Wc: n = h(t;,a) with boundary conditions h(0,0) = h (0,0)=0, h (0,0) = J - cos l ) ) - may be approximated near (0,0,0) as

h(t;,a) = J - cos $1 a + 0(a!) + ]j- sin 4^ aC

jj + 0(3)

The essential features of the unfolding of (2.31) follow by projecting (2.27) on c c the center manifold W : for sin $. * 0 the unfolding on W becomes i = a sin ^ - 2eC - | p C' + iffl C' (2.32)

with A = j 7 + j and where the redundant higher-order terms in a, e and p are dropped o u t . Equation (2.32) shows t h a t t h e p r o j e c t e d s y s t e m u n d e r g o e s codimension-two b i f u r c a t i o n ; however, the physical i n t e r p r e t a t i o n of this bifurcation i s described by the three parameters e, a and p. The neighbouring c o n i c a l s t a g n a t i o n p o i n t s P . ( n . , t ; . ) which l i e on W w i l l depend on these parameters and may be obtained from
u

"i - - I f =i
a sin *! " 2e q - ! u q + Affl q = 0 , i = 1,2,3

Since the C-coordinate of P. s a t i s f i e s a cubic equation, there are one, two or three real solutions depending on the sign of D:
D = - ( c + p*) 1 + (p> + \ e p - a)* (2.33)

-86where we have introduced the shorthand notation a


a sin

-ze

jj_

= ~2i

3 r ^

p =

m m m If D < 0, there are three real and different solutions; for D = 0 again three real solutions appear but at least two of them are equal; finally, if D > 0 one real solution appears and two are conjugate complex. The bifurcation set of the topological saddle point consists of the surface D = 0 bordering domains in RJ where the flow topology is different. A view of this surface is given in Fig. 2.14.

Fig. 2.14. Bifurcation set D = 0; D = (u3 + | - a) z - (6 + u)' The surface D = 0 has two branches which terminate on the curve = -p 2 , a=-J- p' where cross sections of D = 0 with p = constant form a cusp. If u = 0 we observe that a bifurcation solution with three conical stagnation points occurs in a 'wedge' type region near the positive e-axis. For u * 0 this region deformes more or less, so that also three conical stagnation points can occur for an appropriate negative value of e. At the cusp point the two solutions on D = 0 coalesce into the triple solution: S. = p. To obtain the resulting conical flow patterns it suffices to analyze the bifurcation on the center manifold W and to investigate the stability properties of the occurring singularities.

-8 7 The flow on W c which is given by Eq. (2.32) may b e recast i n the form i = f(q;a,,u) = \,((<:-p) s - 3a(C-u) + 2 b )
3. ' '

where a = u* + c and b = a-"izp.-\i' Singular points q = i;. on W

satisfy the relation

(^ - )' - 3a(q - 'M) * 2b = 0;


they are stable if -rz < 0 and unstable i f -rr > 0. dt; dt; For the analysis of the stability properties it is convenient to consider t;. - u as a function of b for different a. The domain of stability is given by |c. - u| < [& so that stability exists only for a > 0. We observe that if three real solutions occur there is one stable solution lying in between two unstable solutions, whereas if one real solution occurs it must be an unstable one. In the special case b = a /a (D = 0) two solutions appear, an unstable hyperbolic point and a nonhyperbolic singularity having a stable and an unstable neighbourhood. In the very special case a = b = 0 a nonhyperbolic unstable singularity occurs. The flow near the center manifold in the cross flow plane may be constructed easily by remembering that the unstable eigenspace E stable singular points on W is empty. It implies that correspond to stable nodes whereas unstable sinHP HP

gular points reveal saddle points in the cross flow plane. c The nonhyperbolic singularity on W having a stable and an unstable neighbourhood represents a saddle-node. In Fig. 2.15 qualitative sketches are given of these cross flow streamline patterns near the center manifold W . The main results about topological saddle point bifurcation may be summarized as follows: - A topological saddle point bifurcates into a stable nodal point and two stable saddle points if D < 0. - If D > 0 the topological saddle point is transformed into a single structurally stable saddle point. - For D = 0 a nongeneric bifurcation occurs which generates a saddle point and a saddle-node. In the very special case D = 0 a n d e = -u 1 , a = -J- p ' (a = b = 0 ) the t o p o l o g i c a l s a d d l e p o i n t r e m a i n s p r e s e n t a n d i s shifted along W respect to its original position. with

-88-

Fig. 2.15. Bifurcation of topological saddle ( j j > 0) bifurcation set (D = 0) and flow patterns near center manifold W . An example of saddle point bifurcation may be encountered in the flow past an elliptic cone at incidence. Cross flow streamline patterns at various incidences were already given in Fig. 2.4. Beyond a particular incidence (say a = a ) the flow structure at the windward side changes qualitatively when the saddle points S. , , . S _ and the nodal point N. are replaced by a single saddle point S. The transition that occurs may be explained as saddle point bifurcation occurring at a = a . The topological saddle point appears at the particular incidence a = a . A small decrease of incidence causes bifurcation into a stable node N. flanked by two saddle points S. and S_. If the incidence is increased beyond a into a saddle point S. Our knowledge about saddle point bifurcation will be applied to show some details of this flow phenomenon. First we note that the flow is symmetric with respect to the minor axis and, moreover, there is always a conical stagnation point present where this axis intersects th body contour. This implies that the flow near the windward side may be described by the system (see Eq. (2.27)) the topological saddle point changes

-89n = vz = ( 2 + 2e) n + j=- (n2 - q2) + vx - + j 2 nq2 + 0(p*) (2.34) v*

q = w z = -2eq - u x nq + ^ i) ;; + u* | ^ - + o(p)
where the bifurcation parameters a and u are omitted. The n-axis coincides with the minor axis of the elliptic body (positive inwards) and the conical stagnation point at the windward side is taken at the origin. Near the origin the body contour will be approximated by n = k. q2 + k_ q* + ... where k and k_ are positive quantities which depend on the principal vertex angles of the elliptic cone. The boundary condition that the body contour coincides with a conical streamline is satisfied if U x = 4^(1-36) + 0(e2) Since e is the only bifurcation parameter, saddle point bifurcation occurs at e = 0. Details of the bifurcation process may be observed clearly from the behaviour of the curves v = 0 and w_ = 0 if E varies near zero. these curves will be given Apart from the symmetry plane (q = 0) where w = 0 by the approximations (see Eq. (2.3*0) vr = 0:
w

n = k1(l-2e) q2 + (k2 + iffl k ^ q' + ...


A

z = 0:

" =2 ^ - (

m-"ki>%

= 0 are sketched in Fig. 2.16 for

The influence of e on the curves v

= 0 and w

A > 0. For e < 0 these curves do not intersect at all, meaning that no conical m stagnation points exist outside the symmetry plane. In the symmetry plane where w = 0 there is a saddle point at the body body surface, corresponding to the flow situation beyond the critical angle a > a

-90-

'_-v r = 0

c): >0

Fig. 2.16. Flow near windward side of elliptic cone.

If E i s increased the curve w = 0 shifts in p o s i t i v e n - d i r e c t i o n and becomes tangent to the curve v = 0 at the origin if c = 0 ; in this case we have to do with the s t r u c t u r a l l y unstable flow at a = a . If e i s increased beyond e = 0, the w = 0 curve moves further and intersects the v_ = 0 curve on the body contour a t C2 = -. Thus for e > 0 two conical
Z A m stagnation points appear at the body surface outside the symmetry plane. 4^5i_jfurcation_of_topological_node

In conical flow a topological node occurs if the cross flow components v

and w

obey conditions similar to those given for the topological saddle point (previous section) except that the sign of A has changed.

The conical streamline pattern near this degenerate singularity may be obtained from Eq. (2.27) as

n = v^. = -2n + ^p (n2 - q 2 ) + 0(p')

q = w z = - u ^ q + v* g- + 0(p')

The detailed structure of the flow depends on the a c t u a l values of u. and v* . The topological node contains two streamlines tending to the stagnation point in Ti-direction and an i n f i n i t e number of s t r e a m l i n e s which approach in the Cdirection.

-91Analogous to the case of the topological saddle point the bifurcation phenomenon is governed by the three parameters a, e and p. Application of the center manifold theory leads to equations exactly the same as those obtained for saddle point bifurcation (Eq. (2.32)). Consequently the results obtained for saddle point bifurcation can be used directly. To interpret the flow on and near the center manifold it suffices to take into account A < 0. Then the following conclusions about nodal point m bifurcation can be established. Nodal point bifurcation in conical flow is governed by the local quantity D (Eq. (2.33)) as follows: - For D < 0, a topological node bifurcates into a saddle-point flanked by two nodal points. - For D > 0 the topological node transforms into a hyperbolic nodal point. - For D = 0 a nongeneric bifurcation occurs generating a nodal point and a structurally unstable saddle-node. In the very special case D = 0 and e = -u2, a = u3 the topological node remains as such and is shifted along the path
v

" = - r ='
A qualitative impression of the flow at and near the center manifold may be obtained by consulting Fig. 2.17.

Fig. 2.17. Bifurcation of topological node ( j i > 0) bifurcation set (D = 0) and flow patterns near center manifold W .

-925. External corner flow; a nonanalytic unfolding of a starlike node*)

51l^_The_flow_around_an_external_corner In the previous paragraph we have discussed the e x i s t e n c e of higher-order conical stagnation points and their bifurcation due to analytical perturbations. In c o n t r a s t to a n a l y t i c a l perturbations also nonanalytical bifurcations may be encountered in conical flow patterns, especially i n flow domains near discont i n u i t i e s in the slope of the contour of a conical body. The e x i s t e n c e of nonanalytical unfoldings may be shown by discussing the supersonic flow around an external corner. An external corner configuration i s composed of two plane delta wings 2. and !_, a t t a c h e d to each other along a common edge such that the planes of the wings include an angle with each other. The two remaining free (leading) edges are supersonic (velocity component normal to the edge i s supersonic) thus the flows on e i t h e r side of the configuration can be considered independently. The flow will be assumed to be conical and the centre of the conical field coincides with the apex of the configuration. W e will be interested particularly in the flow in the region near the e x t e r n a l angle, including both the case of the external axial corner, where the plane wings are nearly perpendicular to each other and the case where the configuration i s similar to one side (upper or lower) of a delta wing with an arrow shaped cross-section (the plane x = 1 in Fig. 2.18). In o r d e r to describe the configuration in more d e t a i l , we use a right-handed cartesian co-ordinate system x,y,z, with the origin in the apex and the positive x - a x i s in the d i r e c t i o n of the oncoming undisturbed flow. The y-axis i s chosen such that the leading edge of 2. i s in the x,y-plane and the z-axis i s perpend i c u l a r to i t . The leading edge of wing Z? and the x-axis determine a plane Q, which makes an angle u with the x,y-plane measured p o s i t i v e as i n d i c a t e d in Figure 2.18. The leading edge of 2. has a sweep angle A. with the y-axis and 2. i s inclined with respect to the x,y-plane at an angle 1 measured in the z , x plane (see Fig. 2.18). The leading edge of Z_ has a sweep angle A_ with the y.zplane and 2_ makes an angle 6_ with Q (<5_ i s measured in a plane through the xa x i s and perpendicular to the plane Q). The line where X.. and _ meet will be called the corner l i n e .

') A substantial part of this paragraph was published by: Bakker & Reyn in the AIAA-Journal, Vol. 23, no. 1, 1985.

-93-

Fig. 2.18. Coordinate systems and geometry of the external axial corner. The conical symmetry of the flow field allows the flow to be described in the cross flow plane Z normal to the corner line. The intersection of 2. and I_ with I are denoted by c r . . and o p respectively, they intersect in the corner point C. The obtuse angle between a- and o_ is called the external angle and denoted by $ ; it can be determined from the parameters 6. , <5?, A . . , A_ and 1 0 . The supersonic flow around an external corner and in particular the existence and character of conical stagnation points in this flow pattern has received much attention in the literature. In the symmetrical case (A. = A_, < 5 . = 6_) the corner point is a conical stagnation point; in the asymmetrical case the possibility that a conical Prandtl-Meyer fan will be formed at the corner point must be considered as well, see Salas and Daywitt (1979). We will restrict the discussion to the case that the corner point is a conical stagnation point. With reference to the local conical stagnation point solutions (paragraph 2 and 3) . which have to satisfy the boundary conditions on o. and cu. various possibilities for the type of conical stagnation points at the corner arise; such as an oblique saddle point or a starlike node. In numerical calculations, made by

-94Kutler and Shankar (1976), both for symmetrical and asymmetrical configurations, an oblique saddle point was found at the corner point and a nodal point on each of the wing surfaces. Furthermore in these calculations the position of the nodal point corresponds to the point where the transverse pressure distribution on the wing surface attains a maximum. A qualitative impression of this flow pattern may be obtained from Fig. 2.19a.

conical streamlines

a) Conical streamline pattern with an oblique saddle point in the corner point (C) and two nodal points IN, and N2)

bowshock
conical streamlines

b) Conical streamline pattern with a single node at the corner point.

Fig. 2.19. Possible conical flow patterns near the corner point of an external corner. In the symmetrical case this flow pattern was also found in the experiments of Bannink (1984). Despite the numerical evidence, Salas and Daywitt (1979) pointed

-95out that a flow pattern as shown in Fig. 2.19b, having a nodal point at the cornerpoint as the only conical stagnation point in the flow field, should also be considered. In fact a numerical study made by Salas (1980) for symmetrical configurations (A. = K = 20, 6. = &~ = 10) and a variation of u from n/2 (external corner type) to n (delta wing type) seems to indicate the occurence of the flow pattern of Fig. 2.19b for e o close to i t . Also, the calculations suggest a transition from the oblique saddle pattern to the starlike node pattern if < o is varied from n/2 to n. In order to clarify this situation we want to investigate whether local conical stagnation point solutions applied to the region near the corner point support the idea of a single node pattern and the transition to the oblique saddle at the corner at some critical value of t o . We note however, that the real nature of the conical stagnation point at the corner point can only be found by solving the full nonlinear boundary value problem without using any approximations such as numerical solutions. This, of course is a hard problem, unlikely to be solved. However, there is one noticeable exception, namely if - = _ = 0, in which case 2. and Z ? are aligned with the uniform flow. The resulting flow is the undisturbed uniform flow, which is represented by a single starlike node at the corner point. We will consider the local corner flow as a perturbation of this flow and the conical stagnation points at and near the corner point as bifurcations of this starlike node. The bifurcating solutions have to satisfy the boundary conditions on 1. and _ and moreover in the limit - ,6p 0 they have to revert to the uniform flow. 5^2i_Boundary conditions and_bifurcation modes The analysis is performed in the cross-flow plane Z, the reference system n, t ; has its origin in the corner point C and the n-axis coincides with 0.. If polar coordinates p , < p are used such that n = p cos <p, Q = P sin < p then the corner boundaries a. and cu in 2 correspond to < p = 0 and < ?= $ the corner point C is given by p = 0. Let a conical stagnation point exist at the corner point and let the conical potential F be expanded near the corner point by F = FQ(1 p n Fn(9) Pm Fm(9) + o(pm)). respectively and

-96Then t h e l o c a l solutions:

F v(<p) = c c o s (n<t> + <| Ti ) n n n' F (<p) = B cos(m<p + X ) m m m F v(<p) = K 6 , n < m <m c

cos(mq> + X ) + Y

m'

cos(n<p + < > | ) + 6 , m =m v

n'

c the boundary

o b t a i n e d i n p a r a g r a p h 2 may be u s e d , which have t o s a t i s f y conditions = 0 at < p = 0 and < p= $ .


3<p
e

With Eqs. (2.6) and (2.7) t h e r e i s obtained

t)i

= X

= k ^- , e

k = 2,3,4....

(2.35)

m = t |- , e or i f

E = 3,'t,5

m x , for any 2 = 3 , 4 , 5 , . . . $ ^ -" e

then

m = m , 6 =0 c m S 2n and because m > n > 1 e

Since e x t e r n a l corner flows are considered for n S $

it follows that > k 2. Obviously, imposing the boundary conditions on a1 and a_ is by itself insufficient to ensure the proper embedding in the surrounding main flow of the conical stagnation point solution near the corner. In fact, the freedom to further specify solutions is already expressed by the possibility, of choosing the values of k and S. in the leading terms of the expansion of the conical potential. The exponent n, occurring in the leading term is illustrated in Fig. 2.20 as a function of $ for various values of k. The solutions for the various choices of k e and 0. will be considered as perturbations on the uniform flow given by F = F which has a starlike node at the corner point. Therefore, the coefficients in the expansion for F, such as e , 8 etc. will be K n 'm relation to each other such that p interpreted as small

parameters tending to zero for the uniform flow. These small parameters vary in F (<p) remains the leading term in the expansion, even if the parameters tend to zero.

-97-

kn $e

k'/C

1
1

^\ | A '/, /. '/

! V k = k>4
\

i \

'l

\1 \
i

/ \ 3 \ /
2 *

'

V!
VVi
\ 1 \ *j> 1 N .

N k'/2
.__t__
nodal point at the corner point (starlike) saddle point at the corner point

Ni*! "-t
l
e

'^V//////A/////7?/

no conical stagnation points at the corner point

3n 25 2

2n

Fig. 2.20. Dependence of n on the external angle for different bifurcation modes (k = 2, 3. 't. . . . ) .

As a result the flow near the corner point is obtained as an unfolding of the starlike node, the streamline pattern of which follows from the streamline equation in polar form: p = (1 + p 2 ) F . pF, p*<V = F by taking F = F :

= "F0.P

P<P = 0

The unfolding of this system which satisfies the boundary conditions on c r - and o 2 , is

= F (n E

cos (n<p) + 0(p

) - p)

(2.36)
Pf> = F Q {-n c n m-1. n+1 p 1 sin (nq>) + 0(p m _ 1 ))

where n and m are given Eq. (2.35)Since n is not necessary an integer it is obvious that the unfolding may be nonanalytic. The coefficient e is the bifurcation parameter with bifurcation value

-98-

e = 0 . S i n c e k can take a l l i n t e g e r values 2 i t appears that various bifurcation modes a r i s e . 5 1 3 il _Bifurcations_of the_starlike node Case k = 2: f i r s t bifurcation mode (n = 1
$ e In the first bifurcation mode, the exponent n can take values 1 < n 2 corresponding to external angles n $ < 2n; the external angle $ = 2n is excluded because the existence of a conical stagnation point requires n > 1. From Eq. (2.35) it follows that m = m
h
c

= 3n-2 for 1 < n x and m = 3n/2 for


3

^ S n $ 2. The location of conical stagnation points may be found from the conditions that dp d < p r r = 0 and -r- = 0 are satisfied simultaneously. For 1 < n < 2 there follows from Eq. (2.36) for small values of e , B as locations for these points: n m C : p = 0 (corner point)

N r N 2 : , x = 0, 9 2 = e.

Pl

= p 2 = (n e^ 2 ""; ^

>0

N3

: ,3 = - e, p 3 = ( n

E n

) 2 - n ; en < 0

where higher order terms in e and B are omitted. n 'm It appears that apart from the conical stagnation point at the corner (C), there are either two stagnation points N. and N_ located on o. and o_ respectively, or there is one point N_, occurring as a free singularity in the flow field, Fig. 2.21a. The points N. and N_ have an infinite number of streamlines tangent to the body surface. The nodal point N_ is situated, to a first approximation, on the bisector of the external angle such that an infinite number of streamlines is tangent to this bisector. Since 1 < n < 2, the conical stagnation point at the corner is an oblique saddle point (this chapter, paragraph 3). The oblique saddle has three separatrices; two of them coincide with a. and a ? , the third with the bisector < p= J -$ .

-99Uniforrn flow
En<0

b) nrZ.^grn

Fig. 2.21. First bifurcation mode of the starlike node (k For n 2, the external angle $

2)

equals n and the body contour at C shows no * 0 (e <J )

discontinuity in the slope. Therefore the classification of first-order singularities (paragraph 3) may be consulted to establish that for e the 'corner' point C becomes a tangent node if e E > 0 and a normal node if

< 0. Furthermore we observe that in this case there are no neighbouring

conical stagnation points tending to C if E 0, Fig. 2.21b. This fundamental difference in bifurcation behaviour between the two cases 1 < n < 2 and n = 2 yields the following noticeable result. Assume that a conical stagnation point of nodal type exists on the surface of a smooth body. Furthermore, let this node be hyperbolic in the sense that analytical perturbations can not affect the topological structure in its neighbourhood. Let the contour of the surface be perturbed slightly so that a discontinuity in its slope is obtained ($ > TI) at the location of the nodal point. This disturbance causes a nonanalytic perturbation of the flow field and the nodal point can not be maintained. In the case of a tangent node it must fall apart

-100i n t o an o b l i q u e s a d d l e p o i n t ( l o c a t e d a t a p o i n t where t h e s l o p e i s discontinuous) accompanied by two tangent nodes on the body surface one on e i t h e r side of the corner point. If a normal node i s attached to the body the node b i f u r c a t e s i n t o an oblique saddle point a t the corner and a nodal point away from the body. This normal node bifurcation looks very similar to the l i f t - o f f phenomena which has been explained e a r l i e r as a saddle-node bifurcation (paragraph 4, section 3). Case k = 3: second bifurcation mode [n = ~-) e In t h e second b i f u r c a t i o n mode (k = 3) . the exponent n can take values 3/2 n 3 corresponding to external angles n S $ 2n. The possible range of n, t o g e t h e r with the boundary conditions (Eq. (2.35)). shows that m satisfies the c o n d i t i o n m 4 n / 3 . Consulting Eq. (2.36) we observe a s t a r l i k e - n o d e bifurcation for 3/2 S n < 2 with the following s i n g u l a r i t i e s : C : p = 0 (corner point)

N r N 3 : <f1 = 0, <P3 = 2/3 3>e. P1 = 9^ =

(n e n ) 2 ~ n ; *n > 0 1

N 2 . N / j : <P2 = 1/3 * e . ^

= * e . P2 = P4 = (-n */'*;

cfl < 0

where higher order terms in e and B are omitted. n m The second bifurcation mode has in common with the first bifurcation mode that new conical stagnation points are generated from the original starlike node. The points N. and Nj. are located at the corner surfaces a. For c and o ? , respectively, < 0 whereas the points N_ and N_ appear as free singularities, Fig. 2.22a. > 0 only the conical stagnation points C, N. and N~ exist and for e the points C, N_ and N^ are present. It appears that all the points N. _ _ ^ are stable nodes; N. and N^ have an infinite number of streamlines tangent to the surface, N_ and N_ have an infinite number of streamlines tangent to the separatrices < p = 1/3 $ and < p = 2/3 $ , respectively. The corner point C is an and < p = 2/3 $ . We conclude that oblique saddle point with four separatrices; two of them coincide with a. and a. whereas the others approach C along < i > = 1/3 $ in the second bifurcation mode for 3/2 S n < 2 a starlike node bifurcates into an oblique saddle point at the corner point flanked by two nodes. One of these

-101-

a)3/2<n<2,^<$e<2n

^W
b)n = 2,$ e =^
conical streamlines

77777y-,

Fig. 2.22. Second bifurcation mode of the starlike node (k = 3) nodes lies on the corner surface (a. or a~), whereas the second node appears as a free singularity in the flow. The case n = 2 corresponds to an external angle $ = 3/2 n. Since n is an integer the classification of first-order singularities (paragraph 3) may be consulted to obtain the flow pattern for various values of e . For small values of e a nodal point is formed at C with an infinite number of

n * streamlines tangent to a1 for E > 0 and tangent to o_ for e < 0, Fig. 2.22b. In c o n t r a s t to the case 3/2 n < 2 we observe t h a t for n = 2 there are no conical stagnation points bifurcating from the s t a r l i k e node. In t h e case 2 < n $ 3 the external angle l i e s in the range n $ < 3/2 ti. From Eq. (2.36) i t follows t h a t t h e r e are no conical s t a g n a t i o n p o i n t s tending towards the corner point C for e 0. For e * 0 the corner point remains a n n

-102s t a r l i k e n o d e . The r a y s < p = 1/3 $ and 2 / 3 $ are to a f i r s t approximation

c o n i c a l s t r e a m l i n e s which d i v i d e the flow f i e l d i n t o t h r e e s e c t o r s . The c o n i c a l s t r e a m l i n e s i n each s e c t o r are curved a t the s i n g u l a r p o i n t , but t h e sign of the c u r v a t u r e i s o p p o s i t e i n adjacent s e c t o r s . The c o r r e s p o n d i n g flow p a t t e r n s a r e sketched i n Fig. 2.22c. This f i g u r e g i v e s a l s o an i m p r e s s i o n of t h e p r e s s u r e d i s t r i b u t i o n n e a r corner which may be obtained from Eq. (2.8) and i s given by the

~T~ (E-) o

F2 = 1 - Y ^ r o

22-2 ( - 2 ( n - l ) e n p n cos (n) + n2 e^ p 2 n " 2 0 ( p 3 )}

1 3 The i s o b a r p a t t e r n shows a saddle bahaviour with s e p a r a t r i c e s at < p = 7 $ , "k $ and ? $ on which p = p . b e o I t may b e n o t e d t h a t i n c o n t r a s t t o t h e f i r s t b i f u r c a t i o n mode, the second b i f u r c a t i o n mode i s not symmetric with r e s p e c t to < p = -z $ . Case k = 4: t h i r d b i f u r c a t i o n mode fn = ) e I n t h e t h i r d b i f u r c a t i o n mode k = 4 , t h e e x p o n e n t n c a n t a k e 2 S n 4 c o r r e s p o n d i n g to e x t e r n a l angles n $ i n e q u a l i t y m > 5n/4. Then Eq. (2.36) r e v e a l s t h a t , a p a r t from t h e c o n i c a l s t a g n a t i o n p o i n t a t f i e l d which tend t o C for E - 0. n For n = 2 , $ = 2n t h e c l a s s i f i c a t i o n Fig. 2.23a. For n > 2 at the corner point there is a starlike node similar to that found in the second bifurcation mode. However, the number of sectors is now four and the rays <? = 1/4 $ , 2/4 $ sketched in Fig. 2.23b. The pressure distribution near the corner, which may be obtained from Eq. (2.8) is given by r-1 T F2 _ ip - 2 and 3/4 $ are the conical streamlines bordering these sectors. The corresponding flow patterns that occur in this bifurcation mode are the c o r n e r p o i n t C, t h e r e a r e no neighbouring c o n i c a l s t a g n a t i o n p o i n t s in the flow

the

values

2n. This p o s s i b l e range of

n, t o g e t h e r with the boundary c o n d i t i o n s , Eq. ( 2 . 3 5 ) . shows t h a t m s a t i s f i e s the

of f i r s t - o r d e r

s i n g u l a r i t i e s may be

a p p l i e d and a s t r e a m l i n e p a t t e r n with a s i n g l e node a t the corner p o i n t r e s u l t s ,

(JH

=1

" F ^ V t"2^-1) n pnC0S {m) + n' En p2n"2 + 0(p

>)

-103-

Uniform

flow

tn<0

<=

E -O

. f\ =0

En>0

/i*e

y.i,
a)n = 2 , * p = 2 n isobars

Fig. 2.23. Third bifurcation mode of the s t a r l i k e node (k = k). The isobar p a t t e r n shows a saddle point behaviour with s e p a r a t r i c e s at < p = 1/8 $ , 3/8 $ , 5/8 $ and 7/8 $ on which p = p . We note in p a r t i c u l a r that the pressure on the corner surface (a- and a) and on the bisector < p = 1/2 $ increases with the distance to the corner point for c > 0 and d e c r e a s e s for
en < 0. n

Case k > 5: higher bifurcation modes fn = k ) e For k > 5 the exponent n satisfies n > 2 and n $

2n. For the conical

streamline pattern again the leading terms in Eq. (2.36) may be used, leading to a starlike node at the corner point, having k sectors. It may be noted that as k >= > the flow pattern resembles more and more that of the uniform flow.

-lofts' 4. _Svmmetrical external corners In order to i l l u s t r a t e the use of the classification of bifurcation modes of the starlike node, the flow around a symmetrical external corner will now be d i s cussed. The symmetry implies that k = even, so that the generation of conical stagnation points from the starlike node only occurs in the f i r s t bifurcation first mode (k = 2) . In order to gain more insight into the question whether new conical stagnation points actually arise or not we will investigate the the local corner flow f i t s within the o v e r a l l flow f i e l d . mode in more d e t a i l ; in particular w e direct our attention to the question h o w The pressure d i s t r i b u t i o n on the body surface and the location of the conical stagnation points as function of freestream Mach number and body geometry will receive special attention.

In this particular case the external angle $ angle A, and < o by

is related to wedge angle 6, sweep

^ -1 f2 tan 6 tan A sin u + (1 - tan2 tan2 A ) cos u - tan2 -i $ = 2n - cos I -. T^-I r;T-T;, 1 l e 1 + tan 2 6 + tan 2 tan 2 A ' where 0 cos it, s i n c e n $ 2 n . e

For the first bifurcation mode the expansion for the conical potential F may be written as (Eqs. (2.6), (2.7)) F = FQ(l + e n p n cos ( n q > ) + T^ p 3 "" 2 cos ( n < p ) + o(p3n"2)) (n e ) J M2 _ v v " o . . ,_ . 2n where Y , , = 777^TV - . 1 < n s 2, $ = . 4 4(2n-l) ' e n The pressure distribution may be obtained using Eq. (2.8)
T-l
Y

(2.37)

fE_] l p ' o

! . Ill 2

M 2[ n * L

E2 p

2n

" 2 . 2(n-l) e n

p n cos (n<p) +

2n T^ e j n + 2 ( n - l ) cos 2

(n<p)] p' 4 ""'* +

o(pkn~k)]

where p and M refer to values in the corner point. *o o Since we are particularly interested in the flow around an external corner with compressive surfaces (6 > 0) (and it will be shown later that embedding of the local corner flow is only possible for c > 0), we will restrict ourselves here

-105to e > 0 in which case a saddle point occurs at the corner and two nodal points
n

_1_
.If the p r e s s u r e d i s t r i b u t i o n i s

on the body surface a t p = p = (n e ) N n written in terms of P N , we obtain T-l fP \ 'D' o


=

T-l , r 2(2-n)p 2n-2 ~T o^pN "

2(n-l) 2-n n n PN p cos

(n<p)

, ,1 . n. , _ -0, -l + o ( p * (2-38)

At the corner point the isobars form a center point and on the body surface saddle points occur. These saddle points correspond to a minimum in the wall pressure, which, to first order, is given by
P

N , T *e " n m , M p p- = l ~ 2 -J o N *^o e

Matching of local corner flow with two-dimensional wedge flows To the order indicated in Eq. (2.37) the expression for the conical potential F contains two free parameters e and M , which may be used to match the local corner flow with the two-dimensional flow found in the region downstream of the supersonic leading edge. The flow in this region is similar to the flow over a wedge and will therefore further be referred to as two-dimensional wedge flow. The matching will be performed by requiring continuity of the velocity (in direction and magnitude) on the ray p = p, the intersection with the body surface of the Mach cone of the two-dimensional wedge flow emanating from the apex of the configuration. We then obtain F = (F - pF ) . tan 6 P (2.39) (1 I=i MS,D) {(1 p) F'p - 2pFFp
+

F) = "-f M^D

at < p = 0, $ and p = p Here 8 is the angle between the direction of the two-dimensional wedge flow and the corner line, and M_ n is the Mach number of the two-dimensional wedge flow. Substitution of Eq. (2.37) into Eq. (2.39) yields

-106nE

n Pn_1

<3n-2) ^

p 3 n _ 3 = [ l - ( n - l ) CR pn -

3(n-l)

T^ p 3 n _ 2 o ( p 2 n ) ] . t a n 8 (2.40)

1=1 wi f . . - , - 2 2 n - 2 _ , # _ l X ;n ^ _,0_ o v _ 4 n - 4 . _,>-',, M 0 [1*U ^ M^){n'^p n-2-2(n -l) c ^ n 0 2n( 3 n-2) V ^r " 4 + <P ] = M2D (2.41) which g i v e s c , M ; a s a r e s u l t a l s o t h e l o c a t i o n of the c o n i c a l s t a g n a t i o n n o p o i n t s and t h e p r e s s u r e d i s t r i b u t i o n may be d e d u c e d . I t may be s e e n from Eq. (2.40) that, for small c , c < 0. that can be obtained as a function of 8 and p from ( 2 . 4 0 ) , and 8 have the same sign which implies t h a t for > 0 and for expanding wedge a n g l e s compressive wedge angles (6 > 0 , 8 > 0 ) , e (6 < 0, 8 < 0 ) , c

For small v a l u e s of 8 t h e equations (2.40) and ( 2 . 4 l ) may be u n c o u p l e d s o t h e unknown p a r a m e t e r c then E i s approximated by tan 8 z -n-l , - , ~n ; np + ( n - l ) p tan 8

which gives the following approximation for the location of the nodal points on the wing surface a. or cu in the first bifurcation mode 1
P

.
j

tan 8 i n-l . x8 1 + n .K~ p tan /

2-n

(2.42)

Since p can be expressed in terms of 8 and M ? n by VM P= - 1 tan 8 + 1 VM| D - 1 - tan 8


in

t h e s h i f t of t h e nodal p o i n t away from t h e c o r n e r p o i n t can be d e t e r m i n e d t e r m s of t h e p h y s i c a l v a r i a b l e s $ , 8, M_ n . F i g u r e 2 . 2 4 shows f o r d i f f e r e n t Mach numbers of the two-dimensional wedge flow (M__ = 1 . 5 . t h e l o c a t i o n of the nodal p o i n t as a function of the e x t e r n a l a n g l e , $

three and t h e

3 and 5)

flow d i r e c t i o n 8. I t may be observed t h a t for e x t e r n a l angles c l o s e to n ( r e p r e s e n t a t i n g c o n f i g u r a t i o n s of d e l t a wing type) the nodal p o i n t l i e s very c l o s e t o t h e corner p o i n t . This s h i f t away from t h e c o r n e r becomes more a p p a r e n t for i n c r e a s i n g Mach numbers and a d i v e r g i n g of the flow towards the l e a d i n g edge.

-107-

8 = 0.04

0.08

0.12

0.16

/
270

240

I//
M2D=1.5 0.02 9=0.04 0.04 0.08 0.06 0.08 0.12 0.10 0.16 0.12 -*rN

210

180

270

(= f ) (= Z) which x x M 2D =3 d e s c r i b e t h e c r o s s flow 210 '/ plane 2 perpendicular to the x-axis. Remark t h a t in the 180 0.02 0.04 0.06 0.08 0.10 0.12 limit 6 o the nodal point 8=0.04 0.08 0.12 0.16 e s h i f t s towards the c o r n e r / / point i f compared with the ' 270 p o s i t i o n of the node t h a t / would appear for the flow 240 around a single wedge, since M2D=5 210 then () = 0. ' ^ Figure 2.25 seems to i n d i 1800 0.02 0.04 0.06 0.08 0.10 0.12 cate that the bifurcation of '-- ^N the s t a r l i k e node i n t o an oblique saddle point and two Pig. 2.24. Location of conical stagnation point nodal p o i n t s i s a h i g h e r on body surface as a function of flow order effect in 6, since for direction (8) and Mach number (M2D) any f i n i t e Mach number M , of the two-dimensional wedge flow. the l i n e s for n = n_ and TI = nN seem to a p p r o a c h t h e o r i g i n a t equal slope. This effect may be verified by expanding Eq. (2.42) near 6 = 0 and taking into account that for < o = n/2, A = 0, tan 8 = s i n 6; then the leading term for PN becomes
240

wv^
IA

1/

If the matching procedure i s applied in t h e c a s e of a s y m m e t r i c a l corner with compressive wedge a n g l e s , c h a r a c t e r i z e d by u = n / 2 , A = 0, the l o c a t i o n of the conical stagnation points on the s u r f a c e s i s found t o depend on the parameters 6 and M^ as shown i n F i g . 2.25. Note t h a t the results are given in the coordinates "

PN = P

1-n 1 2-n 2-n


" 6'

-108-

fan 6
IL

0.28

\ \ \ \

\
0.24

\ \ \
1.5 0.20

'O
\
,MJX
\

\ \ \ \
0.16

\ \ \

0.12

c
N

\
=fan 6

\
\.

0.08

':''

0.0*

-0.24

-0.20

-0.16

-0.12

-0.08

-0.04

x Fig. 2.25. Position of conical stagnation points as a function of free-stream Mach number and wedge angle for a symmetrical configuration wit A = 0, a > = n/2. nodal point n N corner point n

Since on the wedge surface a. we have p = (p - sin 6)(1 + p sin 6) we may write: 1-n 1 * 0(6* ), 1 < n < 2

(cos <5)

L = < 5 + p 2 _ n . 62'n nc = < 5 + 0(6')

-109-

This calculation indicates that in a usual perturbation theory where < 5 is the perturbation parameter, such as used by Vorob'ev and Fedosov (1972), bifurcation of the starlike node is unlikely to appear.

In order to compare the pressure distribution on the wedge surfaces with numerical and experimental results we consider the case of 6 = 10*, M = 3 ; then ti=

2.580 and p = 0.0783- Applying Eq. (2.38) we obtain p/P N which is shown in Fig. 2.26. The agreement with numerical calculations of Salas (1980) and experimental results of Bakker et. al (1981) is quite satisfactory even at greater distance from the corner point. We note that in the numerical calculations entropy gradients are taken into account, whereas the present theory assumes potential flow near the corner, and only a limited number of terms in the expansion are used.

I i

present t h e o r y ( M - = 3 , 6 = 10) G O O o experimenrsfM =2.94,6 = 10.4)

-0.2 C

-0.1

0.0

0.1

0.2

0.3

0.4

t
I cornerpoinr

**

Fig. 2.26. Pressure distribution along the wedge surface of symmetrical external corner with A = 0, u = n/2.

Theoretical observations So far the use of the first bifurcation mode seems a correct way to describe the flow around a symmetrical external corner with compressive surfaces. As a result

-notwo nodal points are found on the body surface as well as a saddle point at the corner. This conclusion is unaffected by a change of ID from < o = n/2 (the corner type configuration) to u i t (the delta wing type). In fact numerical calculations were performed by Salas for symmetrical configurations (6 = 10, A = 20 and 40), M^ = 3 and values of < o ranging from 90 to 180. According to Salas, these calculations indicate that for higher values of u > nodal points on the surface are not present and that the corner point is then a nodal point instead of a saddle point. This would imply that for a certain value of ID a transition from the first bifurcation mode (k = 2) to a higher bifurcation mode (k = 4,6,...) would take place. However such a transition also implies the sudden disappearance of the nodal point situated away from the corner. This is impossible since the nodal point away from the corner is structurally stable with respect to changes in the flow due to variation of 9 . We further remark that if the nodal points are close to the corner point, they are very difficult to detect in numerical calculations, in contrast to the situation when they can be clearly observed further away from the corner point.

In order to show how close the nodal points are t o the corner p o i n t as u tends to 180 for the configuration M = 3 , 6 = 10 at A = 0, 20 and 40, we c a l c u l a t e d the p o s i t i o n of the conical stagnation points according to the f i r s t bifurcation it mode; as shown in Fig. 2.27. For t h e s e c a l c u l a t i o n s appears that for $ < 220 close

t h e n o d a l p o i n t s and the corner p o i n t are so


-0.2*.

*n
Fig. 2.27. The conical stagnation point on oM 3, < 5 = 10 corner Salas (1980); A, O Salas (priv. comm.) node

t h a t a numerical detection would be r a t h e r For $ unlikely. > -220 t h e r e i s a

e substantial deviation of the corner point. If we compare the r e s u l t s of the present

nodal p o i n t s away from the

-Illtheory with the numerical results of Salas (1980) there is a good agreement for A = 40, whereas for A = 0 and 20 the agreement is less. As mentioned before it should be noted that the present results are obtained by using a finite number of terms in a series expansion of potential flow solutions, whereas the numerical calculations use the Euler equations. An experimental observation As experimental results confirming the predicted flow pattern according to the first bifurcation mode have already been reported for the external angle configuration by Bannink (1984), it is of interest to investigate whether the occurrence of nodal points, distinct from the corner point, can also be observed in experiments with a delta wing configuration, even though these points may lie very close to the corner. For this purpose a flow visualization study was made on the upper side of a truncated delta wing with a flat lower surface ( < u = n rad, A = h0\ 6 = 10, $ = 196.5) at H = 3, Re = 2.3 x 10s per cm. Although w e

one should be cautious in drawing conclusions on inviscid flow patterns from flow visualization techniques, experiments on external corners performed by Salas & Daywitt (1979) and Bannink (1984) have shown that the conical nature of the flow is borne out by the oil flow streaklines. The oil streaklines on the upper face of the wing are photographed when the incidence a of the plane lower surface (measured in the symmetry plane) is set at 0. From this picture the angle < | > of the local streamline with the local ray has been carefully measured. The results are collected in
Y (deg wing

Fig. 2.28; they do suggest that nodal points may be distinguished in the conical flow field. The reliability of these results may be judged by comparing them with those obtained for a =
disfance from apex o a 73,5 mm 88,0 mm

-6 = -10, in which case the upper surface is aligned with the flow. Comparing the results for a = 0 and a = 10, a significant shift of the nodal singularity away

x 102,2 mm

Fig. 2.28. Direction of oil flow streaklines.

from the corner is observed.

-112-

6. References
Bakker, P.G. and Bannink, W.J. (197*0 Conical stagnation points in the supersonic flow around slender circular cones at incidence. Delft University of Technology, Report VTH-184. Bakker, P.G. (1977) Conical streamlines and pressure distribution in the vicinity of conical stagnation points in isentropic flow. Delft University of Technology, Report LR-244. Bakker, P.G., Bannink, W.J. and Reyn, J.W. (198l) Potential flow near conical stagnation points, J. Fluid Mech., Vol. 105 Bakker, P.G. (1984) Structural stability and bifurcation in conical flow fields. Delft University of Technology, Report LR-424. Bakker, P.G. and Reyn, J.W. (1985) Conical flow near external axial corners as a bifurcation problem. AIAA-Journal, Vol. 23, No. 1. Bannink, W.J. and Nebbeling C. (1978) Measurements of the supersonic flow field past a slender cone at high angles of attack. AGARD Conference Proceedings 247. Paper 22. Bannink, W.J. (1984) Investigation of the conical flow field about external axial corners. AIAAJournal, Vol. 22, No. 3Bulakh, B.M. (1970) Nonlinear conical flow. Translated from the Russian by J.W. Reyn and W.J. Bannink, Delft University Press (1985). Busemann, A. (1929) Drcke auf kegelfrmige Spitzen by Bewegung mit Uberschallgeschwindigkeit. ZAMM, 9 (6).

-113Ferri, A. (1951) Supersonic flow around circular cones at angles of attack. NACA TR-1045. Fowell, L.R. (1956) Exact and approximate solutions for the supersonic delta wing. J. Aeron. Sci. 23 (8). Germain, P. (19^9) La theorie gnerale des mouvements coniques et ses applications 1'arodynamique supersonique. ONERA, Publ. no. 3 ^ Goldstein, S. & Ward, G.N. (1950) The linearized theory of conical fields in supersonic flow with application to plane airfoils. Aeron. Quart. II. Grossman, B. (1979) Numerical procedure for the computation of irrotational conical flows. AIAAJournal, Vol. 17, No. 8. Kutler, P. and Shankar, V. (1976) Computation of the i n v i s c i d s u p e r s o n i c flow over e x t e r n a l a x i a l corners. Proceedings of the Heat Transfer and Fluid Mechanics I n s t i t u t e , Davis. Calif. Maslen, S. (1952) Supersonic conical flow. NACA TN-2651. Melnik, R.E. (1967) Vortical singularities in conical flow. AIAA-Journal, Vol. 5i No. 4. Reyn, J.W. (i960) Differential geometric considerations on the hodograph transformation for irrotational conical flow. Archive Rat. Mech. Anal. 6(4). Salas, M.D. and Daywitt, J. (1979) Structure of the conical flow field about external axial corners. AIAA-Journal, Vol. 17, No. 1.

-111S a l a s , M.D. (1980) C a r e f u l n u m e r i c a l s t u d y of flow f i e l d s a b o u t s y m m e t r i c a l e x t e r n a l c o r n e r s . AIAA-Journal, Vol. 18, No. 6. Smith. J . H . B . (1972) Remarks on t h e s t r u c t u r e of c o n i c a l flow. Progress i n Aerospace S c i e n c e s , Vol. 12, Pergamon. Stocker, P.M. and Mauger, F.E. (1962) Supersonic flow past cones of general cross-section. J. Fluid Mechanics, Vol. 13 (3). Taylor, G.I. and Macoll, J.W. (1933) The air pressure on a cone moving at high speeds. Proc. Roy. Soc, Ser. A 139Vorob'ev, N.F. and Fedosov, V.P. (1972) Supersonic flow around a dihedral angle (conical case). Izvestlya Akademii Nauk SSR, Mekhanika Zhidkostii i Gaza, No. 5. conical

-115-

Chapter III

Topological aspects of steady viscous flows near plane walls

1. Local solutions of the Navier-Stokes equations


In this chapter the qualitative theory of dynamical systems w i l l be applied to s t e a d y t w o - d i m e n s i o n a l v i s c o u s flows of continua near fixed and moving boundaries. The t o p o l o g i c a l p r o p e r t i e s of these flows may be derived from s o l u t i o n s of the governing d i f f e r e n t i a l equations. Qualitative considerations about these s o l u t i o n s i n the phase plane w i l l be presented with the aim to p r o v i d e i n s i g h t i n t o t h e p h y s i c s of t h e flow p a t t e r n p r i o r to solving appropriate, initial/boundary value problems. The topology of steady viscous flows i s studied on the basis of local solutions of the Navier-Stokes equations (NS) governing incompressible laminar flow. These l o c a l s o l u t i o n s a r e obtained by performing Taylor expansions for the velocity vector field near a point in the flow field. The steady streamline p a t t e r n near t h i s p o i n t i s represented by the t r a j e c t o r i e s of the second-order system x = u(x,y), y = v(x,y) with u,v denoting the velocity components in a c a r t e s i a n reference system x,y. The Taylor expansion of the velocity field i s not r e s t r i c t e d to the neighbourhood of r e g u l a r p o i n t s (u * 0, v * 0) only but may also be performed near separation or attachment points (u = 0, v = 0) because, as Dean (1950) has shown, the NS-equations allow also for analytical solutions near points where u and v vanish simultaneously. Consider the flow pattern that results if an analytic velocity vector field i s expanded up to the N-th order near an a r b i t r a r y point 0, then the t r a j e c t o r y pattern near 0 i s governed by the system (S): N N-i x =u = 2 2 U i=0 j=0 N N-i y =v = 2 2 V i=0 j=0 x y J + 0(N+1) (S)
,J

,J

x y J + 0(N+1)

where 0(N+1) denote higher-order terms of at least order N+l, composed by powers of x and y. The coefficients U. . and V. . are constants which remain undetermined in a local analysis. Since the flow satisfies the continuity equation and

-116the NS-equations, relationships between U. . and V. . exist, reducing the number


iJ i, j

of coefficients that can be chosen independently. A further reduction of unknowns is obtained by partially fulfilling boundary conditions; for example if one studies flow patterns near a wall where no-slip conditions hold. Our main objective now will be to give a unified description of all topologically different flow patterns that will arise near 0 if the remaining coefficients are varied independently. A systematic treatment of this task, being the subject in this chapter, is briefly outlined below and elaborated in more detail for finite N (N S 4) in the sequel. For finite N, we study, instead of the phase portraits of system (S), those of the truncated system (S N ): N N-i x = u = Z Z U. . x yJ i=0 j=0 ,J

(sN)
N N-i y = v = Z I V i=0 j=0 x yJ
,J

and s i n c e (SN) contains only a f i n i t e number of terms the question a r i s e s whether i t g i v e s a proper d e s c r i p t i o n of the l o c a l flow topology in flows governed by the full Navier-Stokes equations. With regard to t h i s question we assume t h a t the important elements of phase p o r t r a i t s are the singular points and the local t r a j e c t o r y p a t t e r n near these points. If singular points (and t h e i r l o c a l t r a j e c t o r y p a t t e r n ) of system (SN) keep t h e i r q u a l i t a t i v e l y c h a r a c t e r even i f higher order terms are added, then the truncated system (SN) suffices to obtain a qualitative description of the l o c a l flow topology. If these s i n g u l a r i t i e s are hyperbolic, the truncated system i s r a t h e r simple b e c a u s e i n t h a t c a s e a l i n e a r expansion (N=l) i s already s u f f i c i e n t for determining the local topology. Moreover, nonhyperbolic s i n g u l a r i t i e s in the flow pattern can be analysed as well using a truncated system (SN) provided that N i s chosen sufficiently l a r g e so t h a t 0(N+1) terms do not disturb the topologies! character of the degenerate singularity as obtained from (S,.). This implies t h a t for a given N a classification of degenerate s i n g u l a r i t i e s , can be established, and each of them represents a local flow topology which can occur in flows that are governed by the full NS-equations.

-117Apart from these relatively simple local flow patterns containing only one singularity, system (S.J enables us also to encounter local flow patterns consisting of a cluster of singular points forming a coherent streamline structure. In order to ascertain that such a c l u s t e r e x i s t s as a l o c a l s o l u t i o n of t h e NSe q u a t i o n s , we s e e k i t a s an u n f o l d i n g of a n o n h y p e r b o l i c o r h i g h e r - o r d e r s i n g u l a r i t y of system (S N ). Let t h i s statement be explained as f o l l o w s . conditions As we know, h i g h e r - o r d e r or d e g e n e r a t e s i n g u l a r i t i e s o f (S N ) , b e i n g l o c a l s o l u t i o n s of the N S - e q u a t i o n s appear under c e r t a i n s p e c i f i e d
J-1J
i iJ

i n v o l v i n g some o f t h e c o e f f i c i e n t s U. . and V. . to be zero. The unfolding of the degeneracy follows by i n s e r t i n g again the terms whose vanishing i n (S N ) cause the appearance of the degenerate s i n g u l a r i t y . did

To obtain the unfolding only lower-order terms have to added and thus the c o r responding unfolding i s a l s o a l o c a l s o l u t i o n of the governing flow equations. This important conclusion enables us to s t a t e that higher-order s i n g u l a r i t i e s of system (S), although they might seem not to be important i n p r a c t i c e because of their rare appearance, are the corner-stones t o o b t a i n l o c a l s o l u t i o n s o f the NS-equations displaying flow patterns having topologies of increased complexity. The method to study higher-order s i n g u l a r i t i e s and t h e i r unfoldings, as outlined so far, i s worked out in the subsequent paragraphs. The s t r u c t u r a l l y s t a b l e flow patterns, among those obtained as bifurcations of a degenerate s i n g u l a r i t y , applications, others are new and are concerned w i t h t o p i c s on laminar are flows i n t e r p r e t e d p h y s i c a l l y ; some of them appear to be very familiar i n aerodynamic l i k e the g e n e s i s o f laminar s e p a r a t i o n b u b b l e s , flow separation near moving w a l l s , interference e f f e c t s in separation and reattachment flows, and formation of asymmetric standing eddies i n the near wake behind a body.

-118-

2. Steady viscous flow near a plane wall, elementary singular points

2 Z 1^Approximate solutions_near_a_glane_wall Consider t h e i n c o m p r e s s i b l e two-dimensional steady laminar flow in the neighbourhood of a plane wall a t r e s t . The flow w i l l be d e s c r i b e d i n a c a r t e s i a n coordinate system x , y such that the x - a x i s coincides with the wall surface; x i s increasing i n downstream d i r e c t i o n . The y - a x i s i s chosen p e r p e n d i c u l a r t o t h e w a l l s u r f a c e such t h a t the flow near the w a l l o c c u r s in the upper halfplane y i 0. The v e l o c i t y components i n t h e x and y d i r e c t i o n w i l l be denoted by u and v r e s p e c t i v e l y while the pressure i s indicated by the symbol p. The governing equations for steady flow are:

continuity equation:

NS-equations conserving momentum: 3u 3u 3p* , u + v + r^ = v V* u 3x 3y 3x (3-2) 3v 3v 3p* , u + v + T * - = v V* v 3x 3y 3y p* denotes the kinematic pressure % v i s the kinematic v i s c o s i t y of the f l u i d P and V2 stands for the well-known Laplace operator. The kinematic pressure can be e l i m i n a t e d from Eq. ( 3 - 2 ) i n order to o b t a i n a dv 3u single equation for the vorticity w = - T, ox oy 3co 3to , ur- + VT-=v7l I D 3x 3y the s o - c a l l e d v o r t i c i t y transport equation. On t h e w a l l s u r f a c e where y = 0 the flow s a t i s f i e s conditions u(x,o) = v(x,o) = 0 the n o - s l i p boundary i0
0 3o

-119If the velocity components u and v are analytic functions of x and y and have to satisfy the continuity equation and the boundary conditions at y = 0 , they can be written u = y (x,y), v = y v(x,y)

where (x,y) and v(x,y) are analytic functions. To describe the viscous flow near an abitrary point on the wall (say at x = 0) let u and v be expanded as u = y^+a^+a^y+a^x'+acxy+agy 2 aye' +agx'y+axy*+a1Qy*) + 0(5) (3-1*)
v = y'(bj+^x+b^+bjjX2+b5xy+b6y') +0(5)

where 0(5) d e n o t e s f i f t h coefficients.

and h i g h e r - o r d e r t e r m s , a . and b . b e i n g

constant equathe

Since the flow i s supposed t o be a s o l u t i o n of ' t h e s t e a d y N a v i e r - S t o k e s t i o n , r e l a t i o n s h i p s between the c o e f f i c i e n t s a. and b . w i l l a p p e a r . The c o n t i n u i t y e q u a t i o n w i l l be s a t i s f i e d following c o n d i t i o n s hold:

(up t o t h i r d - o r d e r t e r m s ) i f

+ 2b

/i ^

3a

+ 2b

(3-5)

+ 3b 3 = 0

2 a g + 3b 5 = 0

a9+4b6=0 The vorticity transport equation is satisfied (to the first-order) if


2a

l\

3a 6

=0 =0 (3-6)

2a ? + a 9

8v (ag + 3a 1Q ) = a ^ Equations (3.*) together with Eqs. (3-5) and (3.6) form a consistent set of relations to describe the viscous flow near an arbitrary point at the wall with fourth-order accuracy. Note that in the given order estimation all coefficients b. can be expressed in terms of the coefficients a..

-120The unknown c o e f f i c i e n t s o c c u r r i n g i n Eq. 3 ^ can be used f o r a p r o p e r of t h e l o c a l flow i n t o a surrounding main flow. The number of t h e s e by E q s . t h a t can be chosen independently i s reduced considerably by the r e l a t i o n s ( 3 - 5 ) and ( 3 - 6 ) . For a f o u r t h - o r d e r expansion, as given i n Eq. {a., a-, a_, a^., a,^, a_ and ag) a r e l e f t i n s t e a d of only seven c o e f f i c i e n t s original sixteen. Our main o b j e c t i v e h e r e w i l l be a d e s c r i p t i o n of a l l t o p o l o g i c a l l y flow p a t t e r n s different appear fitting given {3-^), the

coefficients

t h a t w i l l a r i s e n e a r ( 0 , 0 ) i f t h e s e v e n unknowns a r e v a r i e d

i n d e p e n d e n t l y . Moreover, the c o n d i t i o n s f o r which t h e s e flow p a t t e r n s worthwhile t o r e l a t e the c o e f f i c i e n t s a.

w i l l b e f o r m u l a t e d . To i n t e r p r e t t h e s e c o n d i t i o n s i n a p h y s i c a l c o n t e x t i t i s t o w a l l s h e a r s t r e s s , p r e s s u r e and g r a d i e n t s of w a l l s h e a r s t r e s s and p r e s s u r e . The wall s h e a r s t r e s s i s u s u a l l y d e f i n e d as x = P ( T ~ ) _ n where p i s t h e dynamic v i s c o s i t y of the f l u i d . C o n s u l t i n g Eqs {3.h)


T

and (3-5) one can v e r i f y the r e l a t i o n s :

a. =
1

M
T p

a 2
a_

- -

- u 1 Px 2 u

= + =

2 u

= _

2 u
1 p 1 p
=

"

1 xx 2 u 1

1 *yy " 2 p

i xxx ^ ~ 6 p

xxy _ jL yyy 6 p ~ 6 p xxx 1 p = " 5 xyy p

1 5

which h o l d i n t h e p o i n t ( 0 . 0 ) . I n d i c e s a r e used t o i n d i c a t e p a r t i a l d e r i v a t i v e s . The s t r e a m l i n e p a t t e r n n e a r t h e w a l l i n t h e u p p e r h a l f p l a n e y i 0 may be d e r i v e d from t h e v e l o c i t y f i e l d u ( x , y ) , v ( x , y ) as

-121dx = d = , . ' y)

u(x

(3-8)

y = it

= v(x,y)

t denotes a time-like parameter varying along streamlines. Introducing the Taylor expansion of the velocity components from Eq. (3-'t) into (3.8) we obtain a nonlinear autonomous system for the description of the streamline pattern; the phase plane represents the real flow domain, the phase trajectories may be identified with the streamlines and t serves as a parameter varying along a trajectory. Since y is a common factor in u(x,y) and v(x,y), the line y = 0 is a line of singular points implying a quasi nonhyperbolic singular character at the wall surface. Since the trajectories of system (3-8) are identical with those of the equivalent system -1 x =y u(x,y), -1 y = y v(x,y)

(3-9)

with ( ) = r q j = - -T-, we can remove the singular character of the line y = 0 by investigating Eq. (3.9) instead of Eq. (3.8). In Eq. (3-9), t* is a time-like parameter varying along the trajectories (streamlines), The dynamical system under consideration becomes

x = a^a^+a^y+ajjx'+a^y-^a^+a^x'+agx'y-^a-xy^rt-g^

ag) y'+0(4) (3.10)

y = -^y-a^xy-^y'-^x'y-^agxy'+^y'+OCt) The elementary singular points of this system will be analysed below (section 2.2 and 2.3); the classification of higher-order singularities will follow in a separate paragraph (paragraph 3). 2i2i_Elementary sing^lar_goints_located_at_the_wall_^on-wall_sing^ Let us analyse the flow near a point on the wall where system (3-10) displays a singularity. We take the origin of our reference system in the singular point, then a1 = 0 which means that the wall shear stress vanishes in the singularity; T(0,0) = 0. System (3-10) may now be written:

-122x = a 2 x + a_y + a^x + a^-xy - ^ a^y + 0(3) 1 1 y = - 2 a 2 y " a 4 x y " 3 a5yl

^3^
+ A- = p
a

The eigenvalues of the linear part yield A1

an<

2 a_, thus

i f a 2 * 0 the s i n g u l a r point a t (0,0) i s a (hyperbolic) saddle and the linear part suffices for the d e s c r i p t i o n of the flow topology. The saddle point has an oblique s e p a r a t r i x i n t h e separation - a2 (a2<0) d i r e c t i o n y = - -t x and a separa2a3 t r i x along the wall surface (y = 0 ) . attachment The flow p a t t e r n s t h a t a r i s e for (a 2 >0) various values of a. (a- > 0) a r e shown in Figure 3 - 1 ; for a_ < 0 the flow separates from the wall and for a_ > 0 flow attachment occurs. Fig. 3.1. Oswatitsch-Legendre solution, The flow p a t t e r n for a_ < 0 may be hyperbolic singularity on the o b t a i n e d by t h e t r a n s f o r m a t i o n wall. x -x.
With Eq. (3.7) the separation angle 9 may be written in terms of shear stress gradient and pressure gradient at the separation point as

tan 8

. sep -

3.a2_ 2 a -

3 TX(0.0)

P x (0,0)

3 P y (0,0) P x (0.0)

(3.11)

This classical result was already found by Legendre (1955) and Oswatitsch (1957) and shows t h a t s e p a r a t i o n may be expected in flows with adverse p r e s s u r e gradients (p > 0) and decreasing shear s t r e s s (T < 0 ) . Moreover, Eq. (3.11) shows that small separation angles imply small transverse pressure gradients |p | << |p | , suggesting that a boundary layer approximation y x (p = 0) might s t i l l give reasonable r e s u l t s near s e p a r a t i o n with small separation angles (Van Ingen 1975) ?i2i i e 5 en Ey_i n fi^! a E_E2i n s_in_the_flow_{free_sing^

Returning to Eq. (3-8), free s i n g u l a r i t i e s in plane viscous flow p a t t e r n s are easy t o a n a l y s e . From the continuity equation (3-1) i t follows that (3.8) i s a

-123Hamiltonian system: x = H , y = -H ; H(x,y), the Hamiltonian, is actually equal y x to the familiar stream function *. Streamlines in the flow pattern can be seen as level curves: H = constant on the surface H = H(x,y) in the H,x,y-space. Singular points in the streamline pattern appear if u = v = 0, thus if H = H = 0. The character of a singular point follows in the usual manner from the of the coefficient matrix of the linear part: eigenvalues A ^ ^ / H ' yx \ H xx H yy

-H xy

in the singular point. Then one finds A, + A. = H -H = 0 and A...A- = H H - H . Elementary 1 2 yx xy 1 2 xx yy xy singular points appear if A...A- * 0, they will be hyperbolic saddles if the condition H H -H* < 0 is satisfied in the singularity. xx yy xy In the case H H - H* > 0 the singular point is a center point, at least for wc yy xy

t h e l o c a l l y l i n e a r i z e d system. But the singularity i s also a center point for the nonlinear system since H H - H * > 0 implies that the H(x,y) surface has
xx yy xy a local extremum at the singularity so that the level curves H = constant appear as closed curves, at least in a neighbourhood of the singularity. So we may conclude now that the elementary free singularities either appear as a saddle or as a center. In addition, note that in the special case of an inviscid irrotational flow, where the streamfunction i | i ( x , y ) and thus H(x,y) satisfies Laplace's equation, the eigenvalue product A-,.A_ is negative throughout and consequently only saddle points will appear in such flows.

-124-

3. Higher-order singularities in the flow pattern


Higher-order s i n g u l a r points i n the flow pattern of a viscous fluid may be found i n t h e i n t e r i o r of the flow as well on a w a l l . To c l a s s i f y the degenerate flow patterns that can occur i n a viscous flow i t i s useful to make a c l e a r d i s t i n c t i o n between p o i n t s i n the flow and p o i n t s on t h e w a l l . Let us s t a r t with s i n g u l a r i t i e s i n the flow away from the wall. ^ili.Higher^ordersing^larp^intsinthe^owfield From the c o n t i n u i t y equation (3.1) i t follows that the flow pattern of a s t e a d y 2D viscous f l u i d can be described by a Hamiltonian system x = H , y y = -H

where H(x,y) i s a n a l y t i c and a c t u a l l y equal t o the streamfunction $. Assume that the system has a higher-order singular point a t ( 0 , 0 ) , s o t h a t the conditions H = H x y and ArA2 H H - Hl = 0 xx yy xy

are s a t i s f i e d simultaneously at t h i s point. To study i t s t o p o l o g i c a l properties l e t the Hamiltonian system be expanded as follows

/
yx yy

f
X

\ +

P 2 (x,y) Q 2 (x.y)

-H xx

-H xy

where t h e s e c o n d - o r d e r p a r t i a l d e r i v a t i v e s of H(x,y) are taken i n (0,0) and where P_(x,y) and Q ? (x,y) contain terms not lower than s e c o n d degree and obey the r e l a t i o n

-125Unless H and H are both zero, a nonsingular linear transformation exists xx yy which transforms the system into the equivalent form x = y + P2(x,y) y = Q2(x,y)

and we may apply Andronov's c l a s s i f i c a t i o n scheme for higher-order singular points having both eigenvalues zero (see Chapter I ) . As a r e s u l t we o b t a i n the following degenerate flow patterns:
3'Q2 , * 0
32Q2 32Q2

: cusp point singularity

, ,
ox
3!Q2

= 0,
3Q 2

- * 0: third-order saddle point


oxoy

= .

=0

: flow patterns with a higher degree of degeneracy, t h a t remain unspecified here.

Qualitative sketches of the cusp point and the t h i r d - o r d e r s a d d l e p o i n t are shown in Fig. 3-2.

cusp point

saddle point (third-order)

Fig. 3-2. Higher-order singularities away from the wall. If H and H are both zero, then the Hamiltonian system has a vanishing linear xx yy part in the higher order singular point; the flow patterns have a higher degree of degeneracy and remain undiscussed.

-1263 ; 2 ; _Higher-order singular_goints_on_the_wall In p a r a g r a p h 2 i t was shown t h a t the steady flow p a t t e r n near a wall i s described by an autonomous dynamical system, Eq. ( 3 . 1 0 ) , containing s e v e r a l unknown coefficients (a 1 , a_, a_, . . . e t c . ) . Moreover, if a. = 0 and a_ * 0 this system has a hyperbolic s i n g u l a r i t y at the o r i g i n r e p r e s e n t i n g the ordinary separation or attachment phenomenon. For a_ = 0 the classical Oswatitsch-Legendre solution near separation or attachment f a i l s as a consequence of the nonhyperbolic c h a r a c t e r of the singular point. Moreover,the separation angle predicted by t h i s s o l u t i o n Eq. (3.11) becomes zero. For a_ = 0, a higher-order singularity appears on the wall; a degenerate flow pattern i s formed where shear stress and streamwise shear stress gradient vanish simultaneously. The streamline pattern near such a singular point s a t i s f i e s the nonlinear system x = a-y + a^x' + a,.xy - ^ a^y1 + a_x" + agx'y - ^ a_xy2 - = agy' + 0(4) (3-12) y = - a^xy - x a^-y1 - | a_xly - = agxy2 + ^ a,y' + 0(4)

The shear s t r e s s distribution along the wall near the singular point (0,0) x = u(ajjX! + a-x' + 0(x'))
shows that at the singular point the distribution of the shear stress x attains an extreme value for a^ * 0 and an inflexion point for aj= 0, a_ * 0. For the case a ^ , * 0, the appearance of a shear stress extremum indicates that flow reversal at the wall surface is not to be expected. The flow remains attached to the wall regardless of the occurrence of a particular point where the shear stress and streamwise shear stress gradient vanish. In the case of aj, = 0, a_ * 0 the shear stress changes sign indicating that flow reversal may be expected. The simultaneous occurrence of an inflexion point in the shear stress distribution at the separation point suggests separations of higher-order nature. In order to obtain a catalogue of degenerate flow patterns near a wall, let us analyse the nonlinear system (3-12) near the origin.

-127This system has a nonhyperbolic point at (0,0) with a double zero eigenvalue. To analyse the various possibilities of degenerate flows Eq. (3.12) will be recast into one of the two following forms. A . I a_ * 0; nonzero streamwise pressure gradient x = y + a^x2 + &5xy - | a^y2 + a7x' + &gx*y - i Lxy2 - |figy'+ 0(4) (3-13) y = - a^xy - | a5y* - | a ^ y - | agxy2 + | a?y' + 0(4)

where a. = a. a, B.| a_ = 0; zero streamwise pressure gradient 2 1 1 2 2 2 x = a^x + a^xy - a^y + a_x* + agx y - ^ euxy* - a 3 y ' y = - a^xy - ^ a^-y
2

0(4) (3.14)

- | a_x y - x agxy

+T J a^y' + 0(4)

Note the difference between the two forms: for nonzero pressure gradient (a_ * 0) the system has a linear part whereas for zero pressure gradient (a~ = 0 ) , a system with at l e a s t quadratic terms results. Both forms will be investigated below. Case A: Singularities with a nonzero streamwise pressure gradient (a., * 0) . For a nonzero pressure gradient Eq. (3.13) results which has a l i n e a r part of the form (_ _) . The character of the nonhyperbolic point may be analysed by using Andronov's c l a s s i f i c a t i o n for degenerate points with a double zero eigenvalue (Chapter I ) . O n the isocline of vertical directions (x = 0) y will be expanded as y = tli(x) = k k k A, x + o(x ) where x i s the first nonvanishing term in the expansion

(x) = a^x' a 4 <| a y - | a ^ ) x- * (| a^ + a4(...)) xs o(x)


The expansion for t | > ( x ) points out that higher-order singularities at nonzero pressure gradient may be distinguished, corresponding to a ^ , * 0 and a ^ , = 0.

-128Al: Third-order topological saddle point (a^ 0), x For a^ * 0 we have k = 3, 4 topological saddle point of the third order. The local flow near the third-order saddle point is described by the system
x = y + &J.X1

(0,0) * 0

= 41 so that the degenerate singularity is a

Vy
A s k e t c h of t h e flow p a t t e r n i s shown in F i g u r e 3-3- The s a d d l e p o i n t has four separatrices, two of them coincide with the wall and two s e p a r a t r i c e s a r e tangent to the wall, forming a parabolic curve. If xxx > 0 t h i s parabolic curve l i e s in the lower half plane y < 0 i n d i c a t ing t h a t the flow in the upper half plane remains attached to the wall, see Figure 3-3a. On the other hand i f x < 0, the parabolically curved s e p a r a t r i c e s l i e in the upper half plane so that a flow pattern results as shown in Figure 3-3b.

wall

(b) Txx<0:r

Fig- 3-3- Third-order saddle points,


(p x - ' Txx * 0 ) (a) one hyperbolic sector

(b) three hyperbolic sectors

A2: Fifth-order topological saddle point (&;. = 0 ) . ^ ( 0 . 0 ) = 0 For &r = 0, we observe that the f i r s t nonvanishing term of *(x) i s of the fifth 3 a* if flu * 0. degree (k = 5) having a positive coefficient A. = i According to Andronov's c l a s s i f i c a t i o n scheme, the degenerate s i n g u l a r i t y appears to be a topological saddle point of the fifth order. The local flow near t h i s point i s adequately described by the system
x = y + Lx'

y =

SyX'y

-129The condition aj, = 0 indicates that the second derivative of the shear stress T_ _ _ vanishes at the singularity. The separatrices are found as y= - 4 a_x3+0(x*)

xx 0. and y Corresponding flow patterns are shown in Figure 3 . 4 . The case a_ < 0 represents a flow s e p a r a t i o n phenomenon as s k e t c h e d i n F i g u r e 3 - 4 a . The (a) separating streamline y = - r a_ x' separation (a7<0) l i e s i n t h e upper half plane, leaving the wall t a n g e n t i a l l y i n streamwise direction. (b) attachment The o t h e r case a_ > 0 r e p r e s e n t s (a7>0) 7777777777777777; flow attachement; the separatrix i s * x an a t t a c h m e n t s t r e a m l i n e which terminates a t the attachment point Fig. 3.4. Fifth-order saddle points. tangent to the wall (see Fig. 3-4b). The study of the case a_ = 0, i s not (p * 0, T =0) X XX taken up in t h i s work, a f u r t h e r expansion of the velocity components would then be necessary. Case B: Singularities with zero pressure gradient (a- = 0) Singular flow p a t t e r n s , having zero pressure gradient (a- = 0) in streamwise direction are described by the system (3.14): x = aj.x* + a,-xy - = a^y2 + a^c' + agx'y - - a-xy* - -= agy' + 0(4)
y =

a^xy - - a,-yl - | a_x*y - -z agxy* + -z a_y' + 0(4)

(3.14)

Since the linear part is missing, the classification of Andronov cannot be used in obtaining the degenerate flow patterns. So we are left with the basic question: which terms in the right-hand side of (3.l4) are essential for the local topology of the degenerate state? We will prove that the quadratic part of (3.l4) determines the qualitative pattern provided that aj, * 0. The main tool is a technique called 'blowing-up', for example used by Andronov (1973) and Takens (1974) to study degenerate singularities with at least one linear term. It will be shown here that the blowing-up method also suffices if

-130the l i n e a r p a r t i s missing. The i d e a c o n s i s t s of i n t r o d u c i n g p o l a r transformations coordinate

t o expand d e g e n e r a t e s i n g u l a r p o i n t s i n t o curves c o n t a i n i n g a

f i n i t e number of s i n g u l a r p o i n t s . I f t h e s e a r e h y p e r b o l i c a f t e r blow-up then the l o c a l p h a s e p o r t r a i t n e a r t h e c u r v e , and h e n c e n e a r t h e o r i g i n a l degenerate p o i n t , i s unamenable t o t h e i n f l u e n c e of h i g h e r - o r d e r t e r m s . I n t h e c a s e of Eq. (3-1*0 one blow-up, provided aj, * 0, i s s u f f i c i e n t t o o b t a i n a s t a b l e transformed v e c t o r field. in Here we d e s c r i b e t h i s blow-up i n d e t a i l t o i l l u s t r a t e t h e method. L e t t i n g x = r cos 8, y = r s i n 9 ( r 2 0, 0 8 i n) we g e t Eq. ( 3 - l ' 0 r e c a s t the p o l a r c o o r d i n a t e form 5 1 r = r ( a i , c ' + a_c 2 s - aj.cs 2 - ~ a _ s ' ) + 0 ( r 2 ) (3.15) 8 = -2aj.se 1 - 5 a_cs 2 + 0 a ^ s ' + s . 0 ( r ) where c and s are shorthand n o t a t i o n s for cos 9 and s i n 9 , r e s p e c t i v e l y . We now expand the s i n g u l a r i t y a t r = 0 by r e g a r d i n g ( r , 9 ) a s a c a r t e s i a n d i n a t e system. On t h e i n t e r v a l 0 8 1 1 (3.15) has s i n g u l a r p o i n t s on r = 0 a t 8. = 0 , 9_ = n and 6- j . = t a n (ftj. V3
+

coor-

a 2 ) , where ft- = a^/aj.. p r o v i d e d t h a t aj, * 0 . Moreover, i t appears t h a t the eigenvalue

A s i m p l e e i g e n v a l u e c a l c u l a t i o n shows t h a t a l l t h e s e p o i n t s a r e h y p e r b o l i c

product i s negative saddle p o i n t s . . Q

throughout, are

U'^
Fig- 3 - 5 - F i r s t blow-up of Eq. 3.14 (a,.>0).

implying t h a t these p o i n t s For aj. * 0 t h e ' b l o w - u p ' i s shown i n F i g u r e 3-5 denote flow d i r e c t i o n s S i n c e f o r aj, * 0 a l l

pattern (arrows which

correspond t o the case a^ > 0 ) . singular

p o i n t s a r e h y p e r b o l i c a f t e r the f i r s t blow-up, t h e flow of Eq. (3.15) i s s t a b l e with r e s p e c t t o small ( h i g h e r - o r d e r ) p e r t u r b a t i o n s . C o n s e q u e n t l y , a s i m i l a r conclusion follows f o r Eq. ( 3 . 1 4 ) ; t h e degenerate flow problem i s determined t o p o l o g i c a l l y by t h e q u a d r a t i c terms and i t s t o p o l o g i c a l s t r u c t u r e c a n n o t be d i s t u r b e d by adding h i g h e r - o r d e r terms.

-131Bl: Saddle p o i n t w i t h t h r e e h y p e r b o l i c s e c t o r s (a^ 0 ) . T (0.0) * 0 are

The flow p a t t e r n s t h a t w i l l o c c u r i n t h e p h y s i c a l plane for aj. * 0 may be obt a i n e d i f F i g u r e 3-5 i s 'blown down' from ( r , 8 ) - ( x , y ) . T h e s e p a t t e r n s drawn i n Figure 3-6, the flow d i r e c t i o n corresponds t o an > 0. The b l o w i n g - u p method e v i d e n c e s t h a t t h e t o p o l o g i c a l p r o p e r t i e s p a t t e r n s a r e s u f f i c i e n t l y d e s c r i b e d by the reduced system x = a^x* + a 5 xy - ^ a^y* 1 . y = -a^xy - ^ a ^ 1 The degenerate flow pattern in question has a saddle point behaviour; there are four separatrices, dividing the upper half plane into three hyperbolic sectors. The separation angles a- and ou (Fig. 3.6) satisfy the relation tan a-.tan cu=3of these

tan avtan a2=3 wall

Fig. 3.6. Saddle point with three hyperbolic sectors (p = 0, t


*X XX

* 0). (0,0) = 0

B2: Saddle point with two or four hyperbolic sectors (a;, = 0). T For aj. = 0 the first blow-up given by Eq. (3.15) reduces to = r(a^cls - ^ acs')

(3-16) 4 = -r a^cs* + s.0(r) i 5 On the interval 0 8 S n, (3.16) has three singular points on r = 0, they occur at 9 = 0, 6 = | and 8 = IT. The singular point at 8 = ^ is a hyperbolic saddle point if a, * 0. The points at 8 = 0 and 8 = i t appear to be nonhyperbolic and a second blow-up (r,8) > (n,*) has to be performed. Setting 8 * o , r * 0 we expand (3.1*0 in Taylor series near (r,8) = (0,0) to obtain:

-132r = a r9 + a_r2 + 0(3)


5 = -| ^ a, n -| a59 a?r9 + 0(3)

A similar form is obtained near (r,9) = (0,u). After the second blow-up, defined by 9 = n cos t | > , r = n sin * (n > 0, 0 ty S ^) the following vector field is obtained n = n f a ^ s i n ^ c o s * + a-sin** - ^ a ^ c o s ' * - | a^sin^cos 2 *} + 0(n*) * =I a c sini|icos 2 r(i + a_cosi|isin l * + sini|i.0(n) (3.17)

3 "5*

Within the domain 0 t l > S -r, the s i n g u l a r p o i n t s a r e found a t n = 0,il> = 0 , ^ -1 and ^ = tan ( - 2 / 3 . 8 5 - / 0 , ) . A l l t h e s e p o i n t s a r e h y p e r b o l i c provided t h a t a,. * 0, and eu * 0. Hence the v e c t o r f i e l d of (3-17) i s s t a b l e with r e s p e c t t o h i g h e r order perturbations. We now 'blow-down' once t o g e t p h a s e p o r t r a i t s in t h e r , 9 (Fig. 3 - 7 ) .
O

the

plane 3.7

The f l o w d i r e c t i o n i n F i g u r e

h<Q:

UlLXl
n/2

corresponds t o a_ > 0. The next 'blow-down' t o the p h y s i c a l x , y - p l a n e r e s u l t s in degenerate flow p a t t e r n s as depicted i n Figure 3 . 8 .

UJ|1_J'
a

The blowing-up method shows t h a t the flow t o p o l o g y of t h e s e system (a_ * 0, a_ * 0) x = a_xy a-xy + a_x* degenerate s t a t e s i s f u l l y d e t e r m i n e d by t h e

it/2

1 T

Fig. 3.7. First blow-up of Eq. 3.14 (a,.jj = 0 , a,. * 0 ) .

5Y' "1 VIy

The d e g e n e r a t e flows have a s a d d l e n a t u r e , i f a^a- > 0 (< 0) t h e r e appear two (four) h y p e r b o l i c s e c t o r s i n t h e upper h a l f p l a n e , r e s p e c t i v e l y .

-133As we know, the condition a = aj= 0 corresponds to points at the wall where the streamwise pressure gradient (p ) and the second-order shear stress gradient (T ) vanish simultaneously. This involves that the observed flow may be disand the sort of pressure extreme (maximum cerned according to the sign of T

or minimum). At a local maximum of the wall pressure (a_ < 0) the flow separates

a7>0:

7777777777777.

^77777777777. minimum wall pressure (a5>0)

-7777777777777. maximum wall pressure (a5<0)

Fig. 3-8. Saddle points with two or four hyperbolic sectors


(Px = 0, T ^ - 0 ) .

perpendicularly from the wall surface if T

< 0 (a_ < 0); but there is a more > 0 (a_ > 0) . On the other

complicated separation/attachment structure if T perpendicularly to the wall if T pattern occurs if t


XXX

hand if a local minimum of the wall pressure is present, the flow attaches > 0 and a complicated attachment/ separation < 0.

In conclusion the elementary and higher-order s i n g u l a r i t i e s on t h e w a l l , encountered so far are summarized in the next table.

-134-

singularity

vector field

flow patterns

saddle point Hyperbolic


T(0,0)*0

i,0
Oswatitsch-legendre solution for elementary separation and attachment

-r^^C

a?>0

y*a4x! third-order saddle Non-hyperbolic nonzero pressure gradient


TX(0,0)=0

a\>0

>//?/>////

\-34Xy I

>/?7yy7?;,

px(0.0M0
S <0

'

>f/7f//>/.

fifth-order saddle
5 >0

'

77777777;

saddle point with three hyperbolic sectors Non-hyperbolic zero pressure gradient f* (0.01 = 0 p x (0,0) = 0 saddle point with two or four hyperbolic sectors

x2+asxy-$y2

-xy-}asy 2

a5<0 _ - / f V_ a,<o 777777577


/a5xya,x3

a7>0

'//?}?///,

a5>0 - > 4 ^ a7>o 77777777T. a5>o J a7<0 ' T Z 777:

Table: Hyperbolic and nonhyperbolic singularities on the wall in two-dimensional viscous flow.

-135-

4. Unfolding of the topological saddle point of the third order


4 1 l^_Local_ghase_gortraits_of_the_unfolding The s t e a d y viscous flow along a wall has a t h i r d - o r d e r t o p o l o g i c a l s a d d l e p o i n t if t = T = 0 , x
X
XX

* 0 and p
X

* 0 at this point.

The streamlines form a topological pattern similar to that of the trajectories of the dynamical system (see Eq. (3-13))
2 x = y + a^x2 + &5xy - ^ & / , y +

0(3) (3-18)

y = -a(|xy - | a5y* + 0(3) The degenerate singularity in (0,0) has a double zero eigenvalue implying that c s u the center eigenspace E empty). As a result, the application of center manifold theory provides no reduction of the dimension of the state space which contains all qualitative features of the bifurcation. Consequently we are left with a second-order system whose unfoldings have to be considered in the x,y space. The physical unfoldings, obeying the no-slip boundary condition, follow from Eq. (3-10) by regarding the coefficients a- and a_ as the bifurcation parameters both having zero as the bifurcation value. Small perturbations of a, and a_ with respect to their bifurcation value will be denoted by ]i. , u_ respectively, so that we may write &! = 0 + u ^ & 2 = 0 + p 2 The physical unfolding of Eq. (3-18) becomes . 2 x = ]ix + u 2 x y + ftjjX1 + &5xy - ^ ft^y1 + 0(3) = P ( x , y ; u 1 , w 2 ) y = - 2u2y " a 4 x y " 3 a5yl
+ 0(3) =

coincides with the physical plane (E and E

are both

Q(x.y;n1.u2)

The phase plane portraits of the unfolding are dominated by the singular points (x y ) which might appear after bifurcation. The locations of these points are obtained as those solutions of

-136P(xi,yi;p1>u2) = O (3-19) Q^.y^Uj^u^ = o which satisfy the condition 11m (xi.y) = (0,0) ux 0
U 2 + 0-

The solutions of (3-19) j = x (p-.Pp) and y. = y.(u1,u2)

may be represented by

surfaces. Since p1 and p_ are assumed to be small we are only interested in the local geometry of the surfaces near the origin. This enables us to use a parametrisation technique by means of which the qualitative properties of these surfaces can be fully understood. Using this technique one looks in fact for the leading terms in Eq. (3-19) which determine the position of singular points as functions of the bifurcation parameters . Let x = x]i2, y = y ^ ' P x = k p^, then (3.19) becomes

(k + x + y + & 4 xp p 2 + O(p^) = 0 ( - | y - a 4 x ^ ) p 2 + O(p^) = 0 which shows that the leading terms for x. and y. may be obtained from
u

l + p 2 x i + y i + Vi

y - Voy< - & tlx ^y^ 2p 2yi V i i

These equations may be solved for x. and y. to obtain the following singular points

f-p2 /p
<?

on the wall

l,2p

2ft,, '
2
p

2ft4' W^

away from the wall

-137where p i s a compound bifurcation parameter: p = p* - ^a^p-. Let us c o n s i d e r t h e two cases Sj, > 0 and C U < 0 in the semi-plane y 0 separately, taking a^ > 0 f i r s t . Case &J, > 0: third-order saddle point with one hyperbolic sector For u < 0 no s i n g u l a r p o i n t s occur in t h e semi-plane y i 0; the o r i g i n a l degenerate singularity disappears and the flow remains attached t o the wall. For p > 0 there appear three d i s t i n c t singular points in the upper h a l f p l a n e . The points S. and S_ are attached to the wall and both are saddle points; above the wall and l y i n g midway between these saddle p o i n t s a c e n t e r p o i n t C i s found.
Away from the wall we also observe a saddle-to-saddle connection between S. and Sy enclosing the center point C. A short calculation shows that this connection is given by the equation

= - K ( ( * - ^ > -Sr)
In the particular case u =0 the three singular points collapse into a single point on the wall (x = -u_/2aj,) in which the degenerate singularity, being the original third-order topological saddle point, remains. It means that parameter changes satisfying p consequently p = 0 do not remove nor unfold the original degeneracy; = 0 represents a nongeneric perturbation.

The bifurcation set, together with the possible flow patterns of the unfolding of Eq. (3.18) are shown in Figure 3.9a.

-138-

h
bifurcation set

(a) Unfolding of third-order saddle point with one hyperbolic sector (a 4 >0).

(b) Unfolding of third-order saddle point with three hyperbolic sectors (a t <0).

Fig. 3-9-

Unfolding of the third-order saddle point, bifurcation sets and phase portraits.

-139Case &(, < 0: third-order saddle point with three hyperbolic sectors The unfolding of this degenerate singularity is shown in Figure 3.9b. For u S_ are observed there, resulting in a flow pattern where flow attachement is followed by flow separation (S-). For u < 0 there is only one saddle point in the flow away from the wall and the = 0 represents a nongeneric bifurcation, so that the original local flow structure in the interface between opposing flow directions results. The case u higher-order singularity remains as a structurally unstable element in the flow. > 0 (S.) there appear only singular points on the wall. Two distinct saddle points S.. and

I f ?i_52iEiS_^y^?_5SB5E5^25
Let us consider the unfolding of the third-order saddle point having one hyperbolic sector (aj. > 0) in more detail. A generic perturbation of the third-order saddle point can give, at least locally, either a flow fully attached to the wall or a flow pattern with separation and reattachment. The latter is recognized as the well known two-dimensional laminar separation bubble. Separation bubbles are features in laminar flows that have been studied for many years and it is well-known that near both the separation point and the reattachment point the flow satisfies the Oswatisch-Legendre condition as given by Eq. (3.11). However, in this study the separation bubble also appears in a local solution of the Navier-Stokes equations showing a coherent flow structure including the essential features as separation point, reattachment point, recirculating region, vortex center and dividing streamline. Moreover, this flow structure arises as a bifurcation solution of a nonhyperbolic singularity so that the genesis of separation bubbles can be described in terms of an unfolding of this singularity. A necessary condition for incipient bubble-type separation can now be derived. It is u > 0, or expressed in terms of wall shear stress: > 0

x' - 2t t
X
XX

This condition points out that separation bubbles arise in the flow near a point where T = T
X

= 0 if the flow is disturbed so,that xz - 2TT


' X X

> 0 in this point.

The bifurcation solution for bubble-type separation yields information on the geometry of small separation bubbles. Actually we can calculate the shape of the

-140bubble for the limiting condition u 0. The leading terms for the height of

the bubble h. and for the height of the vortex center h above the wall with D c respect to the bubble length become: bubble height vortex center h./2

t'"c

h / = A IVn c'

Expressed in terms of separation angle 8 we obtain bubble height vortex center h./g = JT tan 9 h / = -? tan 0 c' 6 s in

The first expression shows that bubbles in embryonic state are very slender normal to the wall.

the sense that they are more extended in flow direction than in the direction Furthermore, the center point of the vortex always lies in the upper part of the recirculating region.

In this section we address the problem to describe flow separation near a wall moving in its own plane as an unfolding of the third-order topological saddle point having one hyperbolic sector, ft^ > 0. The set-up of this problem is basically the same as given for the fixed-wall case with the exception that the wall moves with a constant wall velocity u in its own plane. Positive values of u mean that the wall moves in downstream direction whereas negative values of w u represent an upstream moving wall. Separated flow patterns near the moving wall will be investigated by appropriate unfolding the third-order topological saddle point (Eq. 3-13) so that the unfolding satisfies the boundary conditions u = u ,v = 0 w on the wall.

In o r d e r to obtain these unfoldings system (3-10), which i s used in the fixed wall case, has to be replaced by

-141x = u y = + yfaj+apX+a-y+aoc'+aj-xy+a^-y 1 ) y f - ^ y - a ^ x y - ^ 'W*) + 0(4) (3-20)


+ 0

which satisfies the continuity equation (3-1). The vorticity transport equation, Eq. (3-3) requires: u a_
T 1 W 2

2 a + 3a

-*r
=0.

Note that (3.20) reduces to (3.10) if u

Let us examine the phase portraits (flow patterns) of Eq. (3-20) near the origin (0,0), which is a nonhyperbolic third-order saddle point if a. = a. = u also u as a bifurcation parameter having a bifurcation value u = 0 . To obtain flow patterns valid near a moving wall we take, in addition to a- and a_, = 0 . Small = u /a_. variations with respect to this bifurcation value will be denoted by u ing 2 x = u Q + ytuj^x+y+a^x'+^xy - (j ^
u u ? - a^ -^~)

Consulting Eq. (3-20) we find that we have to study the three-parameter unfold-

yz + 0(3)] = P(x,y) (3-21)

y = yM- | u2 - V " 3 S
with a. = a. (a_)

y + 0(3)

=Q(x,y)

and where p , u. and p_ are bifurcation parameters.

Singular points (x.,y.) of (3.21) which appear near (0,0) after bifurcation will be found by using the parametrisation technique adopted in paragraph (4.1) for the fixed wall case. Referring to this case we propose
x

i = V 2 ' yi = h2' Vo = koU2' "l = k l 4

then P(x,y) = 0, Q(xi.yi) = 0

-142becomes
(k

+ k y

i i

+ x y

i i * y' * V j . y i )

u 2

+ 0(p

2)

(-4 - V y^)

M'2 * o(u2)

=o

which shows that the leading terms for x. and y. follow from
P + P

lyi

+ U x y

2 i i

+ y

i + Vi y i "

( P2 + V i } yi =
This system may be solved for x. and y. to obtain two saddle points:

-u fiT
5

i o:
1 f d.

55
'~ih

. :o n

the

a 1 1 i f

= 0 (fixed wall)
^

and one saddle point

, u2 "c " K " u o ( 8 V \


( " 5I~. 55 ) a w a v fronl the wall if u * 0 a **H 4 (moving wall) Furthermore a center point can appear: ^
5 s :

! l

" 24.

'1.

8a

""*

with u = uL - 'fthU-i; the compound bifurcation parameter already encountered in the fixed-wall solution. Note that the bifurcation looks a bit unfamiliar. The free saddle point S, moves to the position (-u2/2&/,, ) o n thew a H i f U + 0 , but this position is not a singular point if u = 0 . On the other hand two adjacent saddle points S. and S- are present on the wall only if uo = 0 ; the singular character of these points is removed if uo is taken nonzero.

-143The unusual bifurcation i s explained if system (3.20) i s understood to unfold a line of s i n g u l a r i t i e s , lying a t y = 0 and appearing in t h e v e c t o r f i e l d x = y1 + a ^ y , y = -a^y'x. The location of the singular points i s e s s e n t i a l l y determined by the compound b i f u r c a t i o n parameter p and the wall velocity p which then appear to describe the unfolding completely. For fixed u and p a v a r i a t i o n of p_ r e s u l t s only in a s h i f t p2/2&j. in xd i r e c t i o n of the s i n g u l a r p o i n t s and thus a s h i f t of t h e c o m p l e t e flow structure. Bifurcation sets and phase p o r t r a i t s in the upper half plane are displayed in Figure 3.10.

'&&//////777;

Fig. 3.10. Flow patterns near moving walls. For p = 0 the fixed-wall case emerges as a particular solution, representing a > 0 and attached flow conditions if p < 0.

bubble-type separation if p

Moving the wall will affect these flows as follows (see Figure 3-10). Let us start with the case that the flow is initially (p = 0 ) separated (p > 0). Moving the wall downstream (p > 0) will at once remove the saddle points Sand S_ and will generate a free saddle point S, and a center point C in the

-Inflow. The closed loop formed by the separatrices above S, is a streamline encircling the closed streamlines around the center point C, forming a region with recirculating flow above the wall. The lower separatrices of S~ extend far upstream and downstream, dividing the flow in an upper and a lower part. The lower part accomodates with the moving wall and flows underneath the recirculating bubble; the upper part of the flow passes over the bubble. A further increase of the wall velocity at constant p (u > 0) will move the center point reaches the critical C and the saddle point S-, closer together, the bubble size diminishes until it shrinks into a single nonhyperbolic singularity when p value P
u

*
(3 22)

o
crit.

lg|")
ft

or expressed in physical terms

i k
uw where p , T, T , T
X X
XX

_
2T

= Q J^ 8u p crit. x |T xx

are to be evaluated in (0,0).

Beyond this c r i t i c a l wall velocity the bubble has disappeared; the flow p a t t e r n adapts s u f f i c i e n t l y to the moving wall and does not contain singular points anymore.
If the flow is initially (u = 0) attached, thus p downstream, p maintained. Consequently, unfolding theory reflects the well-known principle that separation near a wall will be prevented if the wall is moving downstream at a sufficiently high speed. Next, if the wall is moved upstream (p < 0), irrespective of the value of u a flow pattern is observed having a single center point above the wall. This center point is part of a recirculating bubble extending from far upstream to far downstream at least according to the local approximation. To elucidate this large streamwise extension of the bubble let us observe the isoclines x = 0 and y = 0 in the phase portraits of Eq. (3.21). These curves are sketched in Figure 3.11 for the cases p = 0 (fixed wall), p < 0 (upstream moving wall) and p > 0 (downstream moving wall). < 0, and the wall is moved > 0, the unfolding shows that the attached flow conditions are

-145-

Fig. 3-11- Influence of wall velocity on isoclines x = 0 and y = 0. The x = 0 curve has two branches; in the fixed-wall case one branch coincides with the wall surface, the other forms a parabola intersecting the wall at Sj and S_. Imposing a small wall velocity (u < 0) will cause these branches to be perturbed as follows. Apart from a branch in the lower half plane there appears an x = 0 curve in the upper half plane, lying above the unperturbed position and extending from far upstream to far downstream (Figure 3-H)To the order considered, the y = 0 curves (y = 0 and x = -p_/2aj,) remain unaffected by the moving wall. The behaviour of the x = 0 and y = 0 curves in the upper half plane indicate the formation of a recirculating bubble close to the wall surface if the wall is moved upstream. The bubble is small in height but extends from far upstream to far downstream. The previous observations concerning the unfolding of Eq. (3-21) seem to support the idea that: - Moving a wall downstream can prevent or at least delay separation and bubble formation. - Moving a wall upstream will always lead to a 'separation' of the main flow; a small recirculating layer appears underneath the main flow.

The flow patterns along moving walls, obtained by unfolding a third-order saddle point with one hyperbolic sector, will now be used to comment on the so-called MRS-criterion, developed to predict separation in unsteady flow situations.

-146I t i s generally agreed that Prandtl's separation c r i t e r i o n : 3u/3y = 0 at y = 0, i s adequate to predict separation in steady flows. However, i n unsteady flows with moving separation regions, Prandtl's criterion f a i l s and a suggestion was made to replace i t by the so called MRS-criterion developed independently by Moore (1958), Rott (1956) and Sears (1956). The MRScriterion predicts separation in unsteady flows if the condition 3u/3y = u = 0 i s s a t i s f i e d i n some point of the flow f i e l d . The MRS-criterion (3u/3y = 0, u = 0) reduces to Prandtl's criterion (3u/3y = 0, y = 0) i f the wall (y = 0) i s at r e s t in the reference frame in which u i s measured. Points in the flow field where the condition u = 0, 3u/3y = 0 i s f u l f i l l e d are called MRS-points. I t i s not clear, however, what i s to be understood p r e c i s e l y by s e p a r a t i o n in t h i s context. This will be evident in the steady flow along a moving wall, this flow i s unsteady in a reference system fixed t o the w a l l , but no streamline separates from the wall. Using the MRS-criterion (3u/3y = 0, u = 0) Sears and Telionis (1975) proposed a model for separation in the steady flow near a moving wall. This so-called MRSmodel i s p r e s e n t e d i n Figure 3-12, where streamline p a t t e r n s and v e l o c i t y distributions u(y) are shown. Irrespective of whether the wall moves upstream or downstream t h e MRS-model r e v e a l s the formation of a single saddle point in the flow away from the moving wall. The velocity distributions u(y) indicate that this saddle point i s a MRSpoint, satisfying the MRS-criterion 3u/3y = 0, u = 0. Several numerical and experimental attempts have been made to verify the MRSmodel together with the corresponding v e l o c i t y p r o f i l e s by considering the steady flow along a moving wall. A concise and profound review of most of these attempts i s given by Williams (1977). W e do not feel the need to summarize them here i n d e t a i l , apart from some remarks supporting William's conclusion that the MRS-model i s a p p l i c a b l e for the downstream moving case; for upstream moving walls the question i s s t i l l open. Concerning t h e a p p l i c a b i l i t y i n the downstream case the c o n t r i b u t i o n s of Telionis and Werl (1973), Tsahalis and Telionis (1973) and Danberg and Fansler (1975) have t o be mentioned in p a r t i c u l a r . Their r e s u l t s are based on the numerical solution of the boundary layer equations and show a close agreement of the calculated velocity profiles with those expected from the MRS-model.

-147-

Fig. 3-12. Steady streamline patterns and velocity profiles near a moving wall according to Moore Rott and Sears. Furthermore the breakdown of these calculations near a station in the flow where the velocity and the shear stress vanish simultaneously, seems to indicate the existence of an MRS-point in the flow. More recently, Inoue (1981) presented a numerical study of the steady flow along a moving wall. Instead of using the classical boundary layer equations, he solves the Navier Stokes equations approximately by neglecting only the diffusion terms in the main flow direction. This approximate version of the NS- equations has some advantages in comparison with the boundary layer equations. First, the equations can be solved through regions of reversed flow with a prescribed external velocity distribution without meeting any singular behaviour near the separation point (Goldstein singularity). Next, while the steady boundary layer equations do not permit solutions with u = 3u/3y = 0, see Danberg and Fansler (1975). the approximation of Navier-Stokes, as used by Inoue, may permit such solutions because pressure variations in ydirection, being related to transverse diffusion, are allowed.

-148i!i5i_yo?2DS_52^el f r
movin

S_?_Eai225

In t h i s section we continue the discussion on flow patterns along moving walls. A comparison i s made between the MRS-model and the unfolding model, as obtained by unfolding the third-order saddle point (section 4.3). Let us s t a r t with the downstream moving wall. Downstream moving wall Concerning a downstream moving wall Inoue's numerical solution yields flow p a t t e r n s very similar to those obtained by unfolding the third order saddle point (unfolding model). In the unfolding model, a closed r e c i r c u l a t i n g bubble is formed above the wall when separation i s already present a t zero wall velocity, see Figure 3-10. The bubble i s bounded by a single streamline which forms a saddle loop with the saddle point (S_) underneath the bubble. Two stagnation points appear in the flow, the saddle point S_ just mentioned and a center point C in the interior of the bubble region, see Figure 3-13Looking for M R S points in the unfolding model we find two of them, one upstream and one downstream of the bubble; neither of them coincides with S_, as suggested by the MRS-model. These observations confirm a flow pattern as suggested by the numerical solution given by Inoue. Let us calculate the MRS-points (u = 3u/3y = 0) from the unfolding model.
downstream moving wall

The leading terms describing the streamline pattern are (Eq. (3.21))

Fig. 3-13- Flow near downstream moving wall according to unfolding model.

x = u = uQ + j^y + u2xy + y' + fl^x'y y =v = i - g P2y' " (3-23)


ft xy >

4 '

and the unfolding model predicts MRS-points in the upper half plane at

m3i

I" 2^ * 4 \/Pc " 8ft4 *V W

(3 24)

-11,9Based on the unfolding model the following remarks on the existence and s i g nificance of MRS-points are made: - Real solutions for MRS-points occur only if p > 0, indicating that MRS-points may appear in flows along downstream moving w a l l s ; they w i l l not appear in flows along an upstream moving wall. - MRS-points will appear in the downstream case if the wall velocity parameter, p , does not exceed the c r i t i c a l value p = (p /L)1 . This c r i t i c a l crit. value equals the value of p where and above which no r e c i r c u l a t i n g region could be observed in the unfolding model. If the bubble disappears due to an increase in wall velocity the MRS-points will vanish at the same time. - In agreement with Inoue's observation, the unfolding model shows the existence of two MRS-points, one downstream and one upstream of the bubble. However,

I -P 2 p c - VPJ. - u 0 (8a 4 )'


neither of them will coincide with S_: \-zr-, KZ 3 \ 24^ 8^ suggested by the MRS-model. Only in t h e two l i m i t i n g c a s e s , namely zero wall v e l o c i t y (p = 0) and c r i t i c a l wall velocity (p = p ) , MRS-points coincide with saddle p o i n t s crit. in the flow pattern. For zero wall velocity the two MRS-points are located on the wall surface, one of them coincides with the separation point and the other with the attachment point (points Sj and S_ respectively in Figure 3-10). At the c r i t i c a l value of the wall velocity parameter p = p both MRScrit. points, the saddle point (S_) and the center p o i n t (C) coalesce in a s i n g l e point above the wall forming a degenerate point in the flow. Except in some p a r t i c u l a r s i t u a t i o n s w observe t h a t MRS-points cannot be i d e n t i f i e d with s t a g n a t i o n p o i n t s in the flow f i e l d signalling the onset of separation. However, MRS-points may be considered as s i g n p o s t s (precursors) indicating the existence of domains with recirculating flow further downstream. Upstream moving wall The unfolding model p r e d i c t s flow p a t t e r n s near upstream moving walls rather different from those arising in the MRS-context. The MRS-model shows a saddle point (MRS-points) above the wall (Figure 3-12) and a reversed flow region close to t h e w a l l . The unfolding model gives a flow p a t t e r n with a c e n t e r point

-150above the wall. This center point is part of a recirculating flow which extends far in streamwise direction forming a layer underneath the main flow. The center of rotation inside the layer is marked by a stagnation point (center point), as shown in Figure 314. Moreover, no MRS-points are found in the unfolding model. The position of the vortex center (y ) relative to the maximum height of the upstream moving wall layer (y.) will be determined for the flow pattern defined by Eq. (3.23). The height of the layer is found from Fig. 3-11*- Flow near upstream moving wall according to unfolding model the dividing streamline between main flow and the recirculating region. The maximum height yff appears at the vortex position. The limiting value of the height ratio h = y /y. for zero wall velocity (u *0) is found to depend on the sign of u : h = 2/3 h = 1/3 J h = 1/2 if u c > 0 if u c = 0 if u < 0

The differences are related to the limiting fixed wall solutions, namely either bubble type separation (u > 0) degenerate flow (u = 0) or fully attached flow (p < 0 ) . Evidently an upstream moving wall interferes more strongly with separated flows than with attached flows. Finally, we consider the velocity profiles u = u(y) along the wall. Typical profiles, as obtained from numerical calculations (Inoue, 198l),are shown in Figure 3-15 for varying wall shear stress. In the unfolding model the u-velocity near the wall may be approximated by the leading terms

u(x,y)

IV - 5 i H

(&i, > o)

-151where x has to be measured respect to the vortex According to t h i s e q u a t i o n be o b s e r v e d in with three

center.

t y p i c a l v e l o c i t y d i s t r i b u t i o n s can recirculating l a y e r s , each one corresponding to an x-position where the wall s h e a r stress
-0.2 0.0

u Fig. 3'15. Typical velocity profiles for the case of an upstream moving wall according to Inoue (1981).

3y J y=0

c 4;

is either negative, zero or positive. The unfolding model suggests the possibility to have two different types of recirculating layers:

one characterized by positive wall shear stress over the whole length of the layer (u < 0) and a second type in which the wall shear stress changes sign. Comparing the results we find a qualitative agreement of the unfolding model with Inoue's calculations.

-152-

5. Unfolding of the topological saddle point of the fifth order 5iii_Pcrigtionof_the_unfolding In t h i s paragraph we will study the topological saddle point of the fifth-order; in these flows separation or attachment occurs t a n g e n t i a l l y along a w a l l . In such a s e p a r a t i o n (attachment) point T = T = T = 0 , T * 0 and the shear
X XX
XXX

stress behaves approximately as a cubic function near the point. The streamline pattern near such a point can be obtained from Eq. (3.13) by taking &j, = 0, resulting in a system given by the following expression up till third-order terms -1 1 x = y . u = y + a xy + Ux' + SgX2y - j Suxy' 1 - ^ figy1

(3-25)

y = y _ 1 . v = - I &5y' - | a ^ ' y - | figxy1 + | y '


Application of Andronov's classification scheme (Chapter I) shows that, if a_ * 0, the streamline topology near the degenerate point (0,0) is similar to that of the equivalent system x = y + _x' (3-26) y =

.2

2 V'

The degenerate s i n g u l a r i t y as given by Eq. (3.26) describes an attaching flow pattern if a_ > 0 and a separating flow pattern for < 0. In both cases the s e p a r a t i n g s t r e a m l i n e i s e i t h e r approaching or leaving the wall tangentially (Figure 3-16).
attachment ^ ^ ^ > wall

Fig. 3.16. Flow near a fifth-order saddle point.

-153The wall shear s t r e s s distribution near the s e p a r a t i o n (attachment) p o i n t given by T(X,O) = ux' + o ( x ' ) . Let us investigate the p h y s i c a l unfolding of Eq. (3-26) s a t i s f y i n g the flow is

equations together with the no-slip boundary conditions on a fixed wall. The discussion will be r e s t r i c t e d to the case of separation, where a_ < 0 (say

V-D.
The physical unfolding of Eq. (3-26) i s obtained i f t h i s equation i s s u p p l e mented with the lower-order terms occurring in Eq. (3.10); the coefficients of these terms serve as the bifurcation parameters. The physical unfolding of Eq. (3.26) becomes x = ]i^ + u 2 x + y + u_x2 - x = P(x,y) (3.27) 1 y = - 2 u2y " u3xy
+

3 2 x*y

(x,y)

The unfolding, Eq. (3.27), contains three bifurcation parameters, p 1 , u_ and u_ whose effect on the flow topology have to be studied. Characteristic for the unfolded streamline patterns are number, type and location of the singular points near (0,0). There are two sets of such neighbouring points. First we have the singular points on the wall surfaces (on-wall points) located at u . . +u , x + (i-x1 - x' = 0, y =0

The second set describes the off-wall singularities given by 1 3 2 u 2 + p3x - 2 x2 = 0,

y = - j ^ - x(u2 + ]i?x - xJ )

The on-wall singular points follow from a cubic equation, indicating that one, two or three singularities can occur on the wall. If two singularities appear, the bifurcation is nongeneric since one of the singularities is nonhyperbolic. Nonhyperbolicity of a singularity on the wall can also occur in the case that three singularities coincide. This occurs if the bifurcation parameters obey the relations: p_ = -3(^/3)*, ]i. = (p-,/3)'. Then it turns out that no off-wall singularities will arise and the only effect of the bifurcation is a displacement of the fifth-order saddle point along the wall to the location x = x u_ . This nongeneric bifurcation is of no interest and can be ignored without loosing essential features of the unfolding by taking u- = 0 throughout.

-154This may be explained as follows. The nongeneric perturbation u-, = - (u.,)*


1
*
i 5

p. = 27 ( M Q ) ' occurs in the three-dimensional space ( p . . , u_, u_) on a spatial curve which approaches the origin (0,0,0) in the direction of the p_-axis. The nongeneric subspace has dimension one so that the universal unfolding Eq. 3-27 is fully determined by the two parameters u. and \i having a span which intersects the nongeneric subspace transversally at the origin. Then the on-wall singularities are given by A: \ix * p 2 x - x^ = 0, y = 0, i = 1,2,3

whereas the off-wall singularities satisfy

B^B.,: ( [f]

, - U l 2 [f)

), u 2 > 0

The on-wall singularities A. are elementary or third-order saddle points while the off-wall singular points B . . and B_ appear to be a saddle point and a center point, respectively, or a cusp if they coincide. Nongeneric bifurcations of this unfolding occur on the corresponding bifurcation sets:

F " * (3 ) ' B c : v2 = <ul < )

,v2s*

These sets, together with the various resulting flow patterns, are shown in Figure 3.17The generic bifurcation of a f i f t h - o r d e r saddle point w i l l give one of the following three topologically different structurally stable flow patterns. The f i r s t and simplest case consists of a single hyperbolic saddle point at the w a l l , corresponding to the ordinary separation case already described as the classical Oswatitsch-Legendre s o l u t i o n . I t occurs i f u_ < 0 and i f (i. > 0,

^ > 3A"* 2 ^
l J

The second flow pattern looks less familiar; it contains a well-known separation point on the wall but in addition a center point and a saddle point in the flow downstream of the separation point. The upstream separatrices of the saddle point form a streamline which encircles the center point by a closed loop. The flow region behind the separation streamline contains a closed recirculating y l ,P2,* region off the wall surface. It occurs for p_ < 0, p- < -['T-]

-155-

Fig. 3.17. Unfolding of fifth order saddle point, bifurcation sets:


P B : l =

s F

,v2,'

' B c : ^2 = <>i<>-

Finally, the third structurally stable flow pattern shows the familiar laminar separation bubble followed by a secondary separation further downstream. It ,P ,J" u l ,u2,* occurs for u 2 > 0, - ( ^2j < + < (y) . Transitional flow patterns between these three types of flow are described by the nongeneric bifurcations. They correspond to s t r u c t u r a l l y unstable flow p a t t e r n s shown i n Figure 3-17- Again, we d i s t i n g u i s h three different possibilities: 1. B : In the flow region downstream of the separation streamline a nonhyperbolic cusp singularity occurs. The cusp s i n g u l a r i t y may bifurcate i n two ways: i t w i l l f a l l apart into a center point and a saddle point, or i t disappears leaving no singularity in i t s neighbourhood.

-156-

2. B : A third-order saddle point with three hyperbolic s e c t o r s occurs on the wall surface. Bifurcation of the third-order saddle point (see paragraph 4.1) r e s u l t s e i t h e r in two hyperbolic saddles on the wall or in a s i n g l e saddle point above the wall. 3. B : A third-order saddle point with one hyperbolic sector occurs on the wall surface. Bifurcation gives either a laminar separation bubble on the wall, or i t disappears leaving no singularity in i t s neighbourhood. Note t h a t t h i s b i f u r c a t i o n process has already been described i n paragraph 4 of t h i s Chapter. 1 2 i _Bubble_cagturing_by_a_secondary_segaration I t will be attempted now to interpret the unfolding of the fifth-order topologic a l saddle point i n a p r a c t i c a l flow s i t u a t i o n where a separation bubble i s followed by a secondary separation. For this purpose i t i s convenient to observe the v a r i o u s flow patterns i f p_ varies at constant p... I t suffices to take p. < 0 since a l l possible flow patterns will then be encountered and the variation of p_ may be r e s t r i c t e d to go from an appropriate positive value p_ > 3 [ - 2~l to u 2 < 0. These flow p a t t e r n s may be thought to be embedded in the flow field over an a i r f o i l , more in particular they would appear in the process of the capture of a laminar separation bubble by a secondary separation. Figure 3-18. Ul f For p_ > 3(- p - ) we have the s i t u a t i o n as given i n F i g u r e 3 - l 8 a . I f p_ decreases below t h i s v a l u e , the secondary separation points moves in upstream r - 2~J " V the direction, approaching the laminar s e p a r a t i o n bubble. At p- = 3 1 secondary s e p a r a t i o n s t a r t s to interfere with the bubble: the question arises how the bubble will be 'captured' by the upstream moving separation. The unfolding of the f i f t h - o r d e r t o p o l o g i c a l saddle p o i n t suggests (Figure 3.l8b) the following answer: before a final s i t u a t i o n i s reached in which the bubble i s completely lost in the separated flow region (p_ < 0), there appears a transitional s t a g e with an off-wall r e c i r c u l a t i n g region downstream of the s e p a r a t i o n s t r e a m l i n e . The appearance of such a t r a n s i t i o n stage, which i s s t r u c t u r a l l y s t a b l e , i s r e f l e c t e d very w e l l by t h e w a l l s h e a r s t r e s s distribution, Figure 3.l8b. When the secondary separation point S_ meets the r e a r p a r t of the s e p a r a t i o n bubble a t the attachment point A the wall shear s t r e s s reaches a local maximum value T = 0.

-157As long as the recirculating region e x i s t s , the wall shear s t r e s s :


T

= y ( U l + u 2 x - x' + 0(3))

shows two extreme values downstream of the separation point S~; a local minimum and a local maximum. The minimum corresponds to the x-position of the center of the vortex and the maximum to the x-position of the free saddle p o i n t . The r e c i r c u l a t i n g region disappears if the wall shear stress distribution shows an inflexion point where the shear s t r e s s gradient vanishes. Finally, i t may be noted t h a t the surface pressure gradient remains positive in a l l these different flow p a t t e r n s .

(a) Laminar separation bubble and secondary separation downstream

(b) Flow patterns and wall shear stress

Fig. 3.18. Bubble capturing by upstream moving separation.

-158-

6. Unfolding of a saddle point with three hyperbolic sectors in a half plane

Paragraphs 6 and 7 w i l l be devoted t o the c l a s s i f i c a t i o n of u n f o l d i n g s of d e g e n e r a c i e s i n t h e case of zero pressure gradient. Referring to paragraph 3-2 where t h e s e degeneracies are described, we can d i s t i n g u i s h between the c a s e s x * 0 and
T

0 . Each c a s e w i l l be considered i n a separate paragraph. Let us * 0.

s t a r t with the l e s s complicated one, T 6 1 l 1 _Universal_ghysical_unfolding

The corresponding d e g e n e r a t e flow i s governed by t h e q u a d r a t i c p a r t o f Eq. (3.14) (a^ * 0)

x = a^x' + a^-xy - -z a^y'

(3-28)

Vy "3 V
The degenerate flow pattern in the semi-plane y i 0 c o n s i s t s of a saddle point with four separatrices which d i v i d e t h e flow i n t o three hyperbolic satisfying s e c t o r s . Figure 3-19The p h y s i c a l u n f o l d i n g 7777777777777-, Fig. 3 . 1 9 . Degenerate flow at p =0, x *0. the flow equations together with the n o - s l i p boundary c o n d i t i o n may be obtained from Eq. ( 3 . 1 0 ) by t a k i n g a. , a_ and a , as b i f u r c a t i o n parameters, each having zero as the bifurcation value. As u s u a l l y , we i n t r o d u c e the p a r a m e t e r s u. , p_ and u_ to describe small perturbations of a., a_ and a. with respect to t h e i r bifurcation value. Since aj. * 0 we may write

= M

l '&2 =

2'

where a. = a./aj., and the unfolding of Eq. (3.28) becomes x = Vx + u2x + M + x' au: xy

I y' 0(3)
(3-29)

I U2y - xy - i ^ y ' + 0(3)

-159where a is taken in the parameter t. The unfolding contains three bifurcation parameters, but it will be shown that only two of them are really necessary for a universal unfolding; thus the unfolding has codimension two. If the unfolding has codimension two the three-dimensional parameter space (uwhich represents parameter combinaspace) must contain a nongeneric subspace

tions for which the original degeneracy (Eq. (3.28)) is preserved after perturbation. This nongeneric subspace represents flow patterns which are topologically equivalent to that of the degenerate point, apart from inconsequential translations in the x,y-frame. Due to the no-slip boundary conditions point along the wall surface. To find this nongeneric subspace we describe the degenerate flow pattern in a new reference frame x,y with x = x - x , y = y s o that (-x ,o) represents the location of the degenerate point in this new reference frame. Equation (3-28) is then recast into the form it is obvious that only those translations are allowed which will shift the degenerate

x = x* + 2x x + a_x y + x2 + a_xy - 0 y2 (3-30) y = -xQy - y - j ^ y ' Comparing the unfolding of the degenerate point in the original system (Eq. (3-29)) with the representation of the degeneracy in a shifted coordinate system (Eq. (3-30)) it may be concluded that the perturbation which satisfies

l =k

u 2 = k 2xo u3= k a ^

where k is a constant, is nongeneric since it leaves the degenerate flow pattern unaltered apart from a shift along the wall surface. The nongeneric perturbation occurs in the three-dimensional space (uj. Up, u~) on a spatial curve which approaches the origin (0,0,0) in the direction (0, 2, a,.), see Figure 3-20. Consequently, the corresponding nongeneric subspace has dimension one, implying that a universal unfolding can be fully determined

-160by two parameters i f t h e s e two parameters are chosen such t h a t their span in the parameter space intersects the nongeneric subspace transversally at the origin. Since ft_ i s finite the plane u,.=0 s a t i s f i e s the t r a n s v e r s a l i t y c o n d i t i o n , so t h a t the following two-parameter family provides a universal unfolding of Eq. (3.28):

nongenenc subspace

Fig. 3-20. Nongeneric subspace of Eq. 3-29. x = ji 1 + u3y + x* + ft xy - ^ y2 + 0(3)


y = -xy - = y z 0(3)

With suitable rescaling of x, y and t, all possible cases for a,. < 0 can be

reduced to ft. > 0. 6 i 2 i _Bifurcation_sets i _flow_gatterns

To find the bifurcation sets for which the flow pattern i s structurally unstable we seek the singular points near (0,0) on the wall as well as in the flow field. The on-wall s i n g u l a r i t i e s S. ,S_ are given by y = 0, x = J-u. and will occur only i f vi < 0. The off-wall s i n g u l a r i t i e s satisfy
yp

'1,2'

x c

'

where K = V(3+ai)'.
For the generic bifurcations the on-wall singularities S..S- are saddle points, while the of f-wall singularities C.,C_ appear to be a center point (C.) and a saddle point (C_) respectively. 8 The nongeneric bifurcations are specified by p. = 0, and p* 7 P p . = 0 and will be denoted as B and B respectively. This bifurcation set, together with s c the various flow patterns in different regions (I-V) are presented in Figure 3.21. At p- = 0 and p^ > 0 we have a third-order saddle point at the wall (where the points S..S- and C. coalesce) and a single saddle point (C_) above the wall.

-161-

Fig. 3-21. Unfolding of saddle point with three hyperbolic sectors, 4 . . bifurcation sets: B : p = 0, B : u. = ^ V-2u., B : p~ = 2J-U-. Small perturbations due to p1 variations with respect to p. = 0 (u, * 0) give third-order saddle point bifurcation as described earlier in paragraph 4. If y. is negative and u_ > 0 a flow pattern results with a separation bubble at the wall underneath the saddle point C_ (region IV); if u, is taken as positive, only the saddle point C_ remains in the flow above the wall (region V ) .

At B

cusp point bifurcation occurs in the flow, causing either the formation

of a recirculating bubble (region III) in the flow domain in between a separation and an attachment streamline if u_ is increased or a disappearance of the cusp point if ji, decreases (region II).

-162W e now note that the phase portraits in region I I I near B and in region IV near u.. 0 ( u , > 0) a r e not homeomorphic, for the flow in IV possesses a heteroclinic cycle whereas the flow in I I I has a homoclinic cycle. Hence there must be an additional b i f u r c a t i o n which t r a n s f o r m s the phase p o r t r a i t s in I I I and IV into each o t h e r . The saddle C- and the center C. s t i l l exist in the region I I I and IV and thus a global b i f u r cation must take p l a c e , perhaps a s a d d l e - p o i n t - t r i a n g l e formed by saddle-saddle connections (ss) between the saddle points S. , S_ and C_ and surrounding the center point To study t h i s global b i f u r c a t i o n , shown in Figure 3.22, we determine the Hamiltonian of the vector field: >/;///;;;//;/;;;. Fig. 3.22. Global bifurcation of saddlepoint-triangle.
P ^ u 3 y + x + a^xy y = -xy - I ^ y '
+

I y + 0(3)

0(3)

which gives the leading terms of the stream function 4i(x,y) near (0,0): 1>(x,y) = \-[lVx
+

2p3y + 3x' + 2&5xy - y1 + 0(3))

(3-3D

Because t h e saddle points S1 and S_ are on-wall s i n g u l a r i t i e s , the presumed s s connections S.-C- and S--C- have to satisfy the condition t|>(x,y) = 0. A simple c a l c u l a t i o n shows that these conditions are fulfilled i f u_ = K J-p.,. Idle points a t (- V-Uj.0), ( J - u . , 0 ) , then a saddle-point-triangle i s formed by saddle J-P-! and ("*(; 3 ) . and a c e n t e r point in 2 K (-ft,., 3 ) . The ss-connections 5 appear to be s t r a i g h t s t r e a m l i n e s i f the higher-order terms in Eq. (3-31) denoted by 0(3), are not taken into account.

-163The geometry of the s a d d l e - p o i n t - t r i a n g l e depends on the scaled variable ft,., i . e . i t depends on the r a t i o of p (T ) midway between the s a d d l e p o i n t s on the wall. The off-wall saddle point C_ l i e s on the e l l i p s e :

3(f)

= 3

the minor axis of which coincides with the wall o n - w a l l saddle p o i n t s , see Figure 3-23. I t t r i a n g l e , as observed in the unfolding of a sectors, has always acute angles. The v o r t i c i t y triangle as calculated from the expression

segment (length 2) between the implies t h a t t h e s a d d l e - p o i n t saddle with t h r e e h y p e r b o l i c distribution in the saddle-point-

! ' ( l - a ( } f ) _ ^a ( |l , 5 e
is shown in Figure 3-23 as well.

(v., . (X)' . ,5)'J

y i

t "7l
V'

-0.5

V1
\-2

'-Vu

1 . 0

3iMf)"-'A-fl

u = -V, 0 0.5 1 . 0

Fig. 3-23. Location of off-wall saddle point and vorticity contours forft_= p /x = 6".

-164-

7. Unfolding of a saddle point with two or four hyperbolic sectors in a half plane

2;l;_yniversal physical_unfolding The last unfolding we want to study in chapter III is the unfolding of the higher-order singularity occurring in a viscous flow if the streamwise pressure gradient and the shear stress gradient x point. This singularity, and parts of its unfolding, seems to be of particular interest because it describes flow patterns that we meet in several practical situations, such as the flow in the near wake of smooth bodies. The application to these flows will be dealt with in the next paragraphs, but we shall first pay attention to the unfolding itself. vanish simultaneously in the singular

The degenerate singularity we want to unfold here, and which was actually investigated in paragraph 3-2, satisfies the nonlinear system x = a,.xy + a_x*

1 3 y = - a^y* - y'y.

(3-32) a , . * 0, a_ * 0

The corresponding phase p o r t r a i t s i n the semiplane y i 0, s e e Figure 3-8, showed that four d i f f e r e n t flow patterns w i l l a r i s e for various choices of a_ and a_. A s e p a r a t e t r e a t m e n t o f a l l t h e s e four flow patterns i s not necessary because Eq- (3-32) i s invariant for the transformations (a^.y) (-a ,-y) and ( a ^ . y . t ) (-a^-y.-t). Thus i f Eq. (3-32) i s studied both in the upper (y 2 0) as well as i n t h e lower (y 0 ) h a l f p l a n e , i t i s s u f f i c i e n t to discuss the p o s s i b l e unfoldings of Eq. ( 3 . 3 2 ) f o r a t most one c a s e , say a_ < 0 and a_ < 0. The corresponding higher-order s i n g u l a r i t y i s shown i n Figure 324, /^V\
x

yy\
T

it

represents the

flow wall;

s e p a r a t i o n where attains = 0,

pressure

a maximum a t t h e

s u i t a b l e r e s c a l i n g a l l o w s t o take Fig. 3 . 2 4 . Degenerate flow a t p


x

a c = a_ = - 1 . S i m i l a r l y t o former
0 1

=0.

c a s e s , the physical unfolding may be

-165obtained from Eq. (3.10) by talcing the ' l o s t ' parameters (a^.a-.a-.a^) as the bifurcation parameters. All of them have zero as the bifurcation value, and as usual the parameters Uj, p 2 > u_ and pj, w i l l be introduced to denote small perturbations of a. (i = 1 to 4). Then we find find 1 the following physical unfolding:

= u

l * P2X+ 1

3y

+ p

*x' ~ x y " x ' y* 3


+

(3-33)

y = - 2 u2y " p4xy

f"

2x ' y * ^

2i21_Determination_of_codimension The number of relevant parameters determining the unfolding completely can be reduced if the four-dimensional parameter space M: (u. ,u_,u_ .u^.) contains a nongeneric subspace where the o r i g i n a l degeneracy (Eq. (3.32)) i s preserved after perturbation. Such a s i t u a t i o n w i l l occur if the effect of the perturbations i s limited to a shift of the degenerate point along the wall surface. Other p o s s i b i l i t i e s which preserve the degenerate point, such as a rotation in the x,y-plane or a shift off the wall, must be ruled out for t h e i r violation of the n o - s l i p boundary conditions. Let us elaborate the shift along the wall and write Eq. (3-32) with a_ = a_ = -1 in the translated coordinate system x = x - x , y = y

xy
\j \J \j \J

(3.34) y = 2 x Q y + 3xQxy * f "


1 3 y*
+

2y x '

Comparing this result with the unfolding (Eq. (3-33)) we observe that the particular perturbation -kx'
0

"l =
V

-3kx^ 2 =
-kx 3 = 0

" = "3* 0
where k is a constant, is the nongeneric perturbation which leaves the degenerate phase portrait unaffected.

-166T h i s n o n g e n e r i c perturbation i s a one-dimensional subspace of M and approaches the bifurcation point ( 0 , 0 , 0 , 0 ) in t h e n o n g e n e r i c d i r e c t i o n ( 0 , 0 , - 1 , - 3 ) .


Tn

u n f o l d i n g of Eq. (3-32) i s f u l l y determined by three parameters, which have to be chosen in such a way that t h e i r span in M actually i n t e r s e c t s the nongeneric s u b s p a c e t r a n s v e r s a l l y i n the bifurcation point. With regard to the nongeneric d i r e c t i o n , the t r a n s v e r s a l i t y condition w i l l be s a t i s f i e d throughout by t a k i n g Uj. = 0 so that the following three-parameter family provides a codimension-three unfolding of Eq. ( 3 . 3 2 ) . x = ]i + u 2 x + u.y - xy - x' + 0(3) 1 y = - 2 u2y 1
+

3 y'

3 2 x'y

(3-35)
+

*3' and_B

2 1 3 i _Neighbouring singular golnts _local_bifucation_sets_B

To o b t a i n i n s i g h t i n t o t h e v a r i o u s p r o p e r t i e s of the phase p o r t r a i t s of Eq. (3-35) we seek the singular p o i n t s , x . = ( x . , y . ) , near (0,0) on the wall as well as i n t h e flow f i e l d . Values of the parameters for which these singular points are n o n h y p e r b o l i c are the b i f u r c a t i o n s e t s a t which the flow p a t t e r n s are s t r u c t u r a l l y u n s t a b l e . S i n c e we o n l y d e a l with phase portraits close t o the origin ( x , y ) = ( 0 , 0 ) we are o n l y i n t e r e s t e d i n t h e l o c a l geometry o f = x . ( p ,p_,p_) and y. = y ^ p ^ . p ^ p , ) = 0.
allows to near tne

the

surfaces x

bifurcation point p 1 =

p_ = p . = 0 where x

The parametrisation x. = x . p ' , y. = VJIU-I ) ' . P? = ^P^l'' u 3 = ^ j ^ l ' ' find these surfaces and phase p o r t r a i t s from the truncated system

Ux + u 2 x + u_y - xy

(3.36)
1 1 H J 2 3 2 5 p,y 5 y
1 +

2 1 o x y

The on-wall s i n g u l a r i t i e s have x - l o c a t i o n s on the wall found as solutions of the cubic equation p. u_x - x' = 0. This equation points out that the number of on-wall s i n g u l a r i t i e s depends on u. and p _ ; a c t u a l l y t h r e e s i n g u l a r i t i e s appear i f ( p 2 / 3 ) ' > ( u j / 2 ) 1 whereas one s i n g u l a r point may be found i f ( u 2 / 3 ) ' < {]x/2)'. hyperbolic. The condition If ( p 2 / 3 ) ' * (Pj/2) 1 i t may be v e r i f i e d t h a t t h e s i n g u l a r p o i n t s are saddle points and that a l l of them are

-167B g : (u-,/3)' = ( U l / 2 ) ' forms a bifurcation s e t , called B . s At the b i f u r c a t i o n s e t B , there will be at most two on-wall s i n g u l a r i t i e s of which at least one i s nonhyperbolic. The type of the nonhyperbolic p o i n t i s established as follows. Let us denote the x-position of the nonhyperbolic p o i n t by x- and t h a t of the remaining saddle point by x_. Since x. i s at least a double root of the cubic equation Pj + u_x - x' = 0 we find easily that x_ = -2x. and x. = -(u.,/2) 1/3 ' . If (x.,0) i s taken as a new origin in the phase plane, system (3-36) becomes 1/3 * = (H3
+

1/3 ) y
+

ij )

3 [j ]

( x - X l ) ' - (x- X l ) y - (x- Xl )> (3-37)

p 1/3 y = -3 (2-) (x-x x ) y + g y' * 2 ( x " x i ) ' y Unless (u_,u,) = (3(u.,/2) , -(u.,/2) ' ) , we can use Andronov's theorems 66 and 67 to establish that (x ,0) i s a third-order saddle point i f p. * 0 or a f i f t h order saddle point if \i. = 0. The particular case: (u 2 > p,) = ( 3 ( ^ / 2 ) ' 3 , - ( U j / 2 ) 1 ' 3 ) , being element of B s w i l l be l a b e l l e d B-. I t describes a nonhyperbolic saddle p o i n t with three 9 1/3 separatrices having the angular coefficients 0, - 5 (u.,/2) and . F i n a l l y we conclude t h a t the b i f u r c a t i o n s e t B \B.. i n d i c a t e s saddle point bifurcations on the w a l l : of t h i r d order for p. * 0 and of f i f t h order for u- = 0. Both types of bifurcation have already been discussed (paragraph 4 and 5) so that the results can be applied here without further comments. The p o s i t i o n s of off-wall singular points of Eq. (3.36) follow from the relations ; 7x' - 9u 3 x l - u2x + 2ux + 3 u 2 u 3 - 0, y - I (u 2 - 3x') (3-38)
+

The x-component s a t i s f i e s a cubic equation, with the result that i f ( 7 l / 3 ' U-i 7/3 U3 u') * (u + 7/3' \i2)' o n e o r t n r e e hyperbolic points appear. 7/3 P P2 " v\)* 2Po The condition

V <V vl + 3 ^ 3 ' W

=(

"3 + U2>'

-168-

forms a bifurcation set indicating the existence of nonhyperbolic points in the flow field. Let us examine the topological type of this nonhyperbolic point in order to determine the character of B . c
If

(xj.y^ is a double root of Eq. (3-38), x ^ ^ satisfies the relation 21x -p = 0 and the linear part of the vector field (Eq. (3.36)): '-" y 9xy 3(u--x)

lSp^

Df = |

'3
y

has zero as a double eigenvalue in (x.,y.), The linear transformation 3 0 3(U3-Xl)

^yl

which i s n o n s i n g u l a r on B \B- allows Eq. (3-36) t o be b r o u g h t i n t o t h e n o r m a l form 2


y

x = y - (i

9 Pj-x^

x*

xy . U3" x i

y = Z7"C O u i - l Z y ^ x . + u - ) x* + 5 xy + ^ T - r y2 + | x ' y + \ y - x ' K J 6(11-^) WK3 3 1 K2' 9 Uo-Xj 3 3(u3-x1) J 2 6 Jl where x = x - x x and y = - x y 1 ( x - x 1 ) + ( u _ - x 1 ) ( y - y 1 ) . U s i n g A n d r o n o v ' s t h e o r e m s 66 and 67 i t f o l l o w s t h a t the nonhyperbolic p o i n t ( x - . y . ) i s a cusp p o i n t s i n g u l a r i t y p r o v i d e d t h a t t h e y - e q u a t i o n h a s a n o n v a n i s h i n g f i n i t e x 2 term, thus i f 9P - 12u_x.. + u_ * 0 . This c o n d i t i o n i s f u l f i l l e d except f o r t h e c a s e u- = u_ = 0 and f o r t h e l a b e l l e d B_ r e p r e s e n t e d by the parameter combinations: case

[2.u3]

[-3 x/g i 2 / 3 . j y? (^)V3i. B2C ,c

For u. = p ? = 0 there occurs a fifth-order saddle point that coincides with a singularity on the wall whereas B- represents a third-order saddle point in the half plane below the wall.

-169In conclusion: the bifurcation set B \(B. u B_) corresponds to cusp point bifurcation in the flow off the wall. In paragraph 5-1 cusp point bifurcations have already been discussed and we have seen there that they cause the spontaneous generation of singular points in regular domains of flow fields. Since the vector field (Eq. (3-36)) i s Hamiltonian, these new s i n g u l a r i t i e s will always appear as a center-saddle pair. 71^ti_Flow P.atterns_and_global bifuation_sets_B and_B

The bifurcation sets B and B divide the parameter space (p..,Pp,p_) in several subdomains where flow patterns of different topology may occur. Standard subdomain methods, such as parametrization and singular-point analysis (c.f. previous paragraphs) may be used to establish these patterns in e i t h e r separately. The task to obtain a clear three-dimensional view of the parameter space and of i t s p a r t i t i o n by B and B in subdomains i s not an easy one; especially the geometry of B is rather complicated. Therefore, w e investigate the case p.. = 0 first and then the influence of u- * 0 will be considered. For p. = 0 Eq. (3-36) reduces to x = p x + u_y - xy - x1 (3-39)
1

p_y M y + - + \ x'j
2

The bifurcation set B : (p-/3)' = (pj/2)2 which concerns the on-wall singularities, reduces for u1 = 0 to the simple form u_ = 0 and coincides with the praxis. However, if p. = 0 the p.-axis also forms part of the B bifurcation set. l 3 c For p.. = 0 the bifurcation set B becomes 1 c (|P2P3

- p' 3 )' = ( P 3 j r n j l '

which consists of three branches p 2 = 0, pi = 0,00334 p 2 and pi = 0.95963 P2A sketch of B and B in the p_,p_-plane is shown in Figure 3- 2 5. This figure s c 2 i shows also the flow patterns caused by unfolding the degenerate singularity near the bifurcation point (p2>p,) = (0,0). I t suffices mirror reflection with respect to the y-axis. to take p, > 0 only. Flow patterns corresponding to p_ < 0 can be obtained from those at p_ > 0 by a

-170-

Fig. 3.25. Partial unfolding of Eq. 3.36 (^ = 0, u.. > 0) bifurcation sets: B g : u2=0, B c : u3=0.9796 /p 2 . u3=0.0578 Ju2. Bg]_: Vy0, \iyh2-

Let us describe the various flow pattern if u ? increases at constant u > 0. In region I we find two hyperbolic saddle points, one on the wall and the other in the lower semi-plane (y < 0 ) . Going to region Ila we approach the line u_ = 0 where the saddle point on the wall becomes a degenerate fifth-order saddle point. Crossing the line p_ = 0 will unfold this degenerate saddle point as described in paragraph 5. Figure 3-17. whereas the hyperbolic saddle in the lower half plane remains qualitatively unaffected. Consequently, the flow region in Ila contains two separation bubbles, one on either side of the wall, and a 'free' saddle point below the wall. In region lib, we find that the n-wall separation bubble in the lower half plane (present in Ila) does not occur but is replaced by a center-saddle pair below the wall. Obviously there is a transition from Ila to lib (and vice versa) which is not predicted by B or B . Detailed calculations show us that this transition may be

-171characterized as the formation and breaking up of a separatrix triangle formed by two on-wall saddle points, the 'free' saddle point in the lower half plane and their connecting streamlines. Since number and type of singular points remain unchanged during this transition, the process will be a global bifurcation. The corresponding global bifurcation set is called B . , and follows gl by assuming that saddle connections exist between the saddle in the flow and the two saddles on the wall. These connections will appear if the condition p. + UpU- - p i = 0 h o l d s .

The t e r m i n a t i n g s a d d l e p o i n t s ( 0 , 0 ) , (u^, 0) and (p.,, 4 (p hyperbolic. The condition:

- 3p?,))

are

V
For p 3-25-

> i + Wi - u3 =
= 0 , B . reduces to the curves p., = 0 and p_ = /pT as depicted in Figure

forms the global bifurcation set B ..

Going from region II (in (Figure 3-25) to region III a B -curve is crossed indicating cusp point bifurcation in the lower half plane; below this curve the center-saddle pair in the lower half plane has disappeared. Entering region IV from III, again a branch of B is passed on which a cusp in the upper half plane bifurcates into a center-saddle pair.

As a result the flow in region IV i s c h a r a c t e r i z e d by an on-wall s e p a r a t i o n bubble together with a center-saddle pair above the wall. On B . where p_ = 0 and p_ > 0 the flow i s structurally unstable and consists of two c o u n t e r - r o t a t i n g v o r t i c e s , surrounded each by three saddle connections in the upper half plane. Passing B . (p_ = 0 ) to p.. < 0 will reverse the mutual xSl 3 J p o s i t i o n s of the on-wall separation bubble and the center-saddle pair as they appear in region VI. Proceeding now to the more general case y- * 0 we consider f i r s t the influence of \i. on the bifurcation s e t s and then on the phase p o r t r a i t s . For nonvanishing 2/3 1/3 p. we may use the parametrisation p_ = k(p./2) , p- = (p./2) J to obtain a unified picture of the bifurcation sets in a plane p. = constant. Due to p. * 0 the b i f u r c a t i o n s e t s w i l l become p e r t u r b e d , as i s shown in Figures 3-26 and 3-27.

-172The parametrized form of these sets becomes for u- * 0: Bs :k =3

B : {98+63 k - 27 V } 2 = {9! + \ k}> B .: e1 - k - 2 = 0 gl Two special parameter combinations to be denoted by B. and B_, appear at (k,) res ectivel

(3.-D and (-3 ^/Is' 3 V F '

y-

The bifurcation sets B and B . are tangent to B in B. . Furthermore B_ is a c gl s 1 2 cusp point of B . c The bifurcation sets divide the parameter plane (u-,u,) in several domains (labelled I, I', II, II', III, III', IV, IV', V and VI) where topologically different phase portraits occur. These phase portraits are obtained by using standard methods such as local linearisation and singular-point analysis. Since these methods have been frequently H,= 0:. , h >0:. discussed in previous paragraphs we omit detailed calculations here and we confine ourselves to stating the results. Fig. 3-26. Bifurcation set B The influence of a nonvanishing \i. on the qualitative properties of B , B and B . is mainly concentrated around the u^-axis. More precisely, beyond the interval p ? (B_) < p_ < u?(B.) they approach, for increasing |up|, more closely to the form they have in the case u1 = 0 . This means that corresponding domains (equally labelled in Figure 3-27) at p. 0 and u . . * 0 will represent similar (topologically equivalent) phase portraits of Eq. (3-36). However, if u p is sufficiently small, the influence of u. is more severe and gives rise to the appearance of two new domains: V (near B_) and VI (near B.) for u1 * 0. The phase portraits in V and VI will be deduced from those found in adjacent domains and by passing the separating bifurcation set. In this way it follows that the flow in region VI is modelled by a hyperbolic saddle point on the wall, a center-saddle pair above the wall and a free saddle point in the lower half plane. For region V more discussion is needed.

gl'

-173-

Fig, 3.27. Bifurcation sets B

and B .

Region V i s subdivided by B into Va and Vb (for details see Figure 3-28) . The flow in Va i s obtained by applying a saddle point bifurcation in the flow in H a . Then i t i s found that a hyperbolic saddle point S1 , occurs on the w a l l , together with a c e n t e r - s a d d l e p a i r (C-S-) and a hyperbolic saddle point S_, both in the lower half p l a n e ; s t r e a m l i n e s e n t e r i n g the region between the c e n t e r - s a d d l e p a i r and the saddle point S_ are prevented from crossing by the separatrix emanating from S 1 . The flow p a t t e r n in region Vb merits further consideration. W e can obtain t h i s pattern from the pattern in Va by crossing the global b i f u r c a t i o n s e t B . , the effect of which will be that (OS-) and S- will now be found on the same side of the separatrix through S. . On B . there will now be a saddle-connection between S. and S_, i n d i c a t i n g structural i n s t a b i l i t y in the global sense. This flow i s obtained as well when going through the upper branch of B from I ' to Vb. However, d e r i v i n g the flow pattern from the pattern in l i b by crossing B (or alternatively from I ' , by crossing the lower branch of B ) , we obtain a p a t t e r n in which (C-S-) forms a center-saddle pair, instead of (C-S 2 ). Consequently, two

-174topologically different flow patterns seem to occur in region Vb. This discrepancy can be removed by assuming a second global bifurcation set, called B _,
g2

which d i v i d e s Vb i n t o the subdomains Vbl and Vb2 such that the center-saddle pairs (C-S..) and (C-Sp) occur in Vbl and Vb2,respectively, see Figure 3.28.

Fig. 3-28. Bifurcations near B_. The bifurcation set B _ represents a flow pattern with two saddle-saddle connect i o n s forming a h e t e r o c l i n i c cycle around the center point (C). To confirm the existence of such a global bifurcation s e t we have to look for a p a r t i c u l a r parameter s e t u_ = p.fji. ,p_) for which the saddle points (S?) and (S_) l i e on the same level curve of the Hamiltonian of the corresponding vector f i e l d . Numerical c a l c u l a t i o n s have been performed to approximate B _; the results are 2/3 1/3 given in the parametrized form u_ = k(u./2) , u_ = {]i./2) where k and <L are given i n the next table. The results point out that B _ i s found in region Vb only and that i t terminates at B_ and at Bj.:the i n t e r s e c t i o n of B with the upper branch of B .

-175-

B2:

-1.963 -I.902 -I.747 -1.2*5 - .657 0.0 .O85 .223 1.217 2.369 3.000

.713 .729 .781 .914 1.078 1.259 1.283 1.317 1.558 1.864 2.000

<V

Table: Numerical approximation of B _. All the previous results concerning u. * 0 may be brought together to obtain a total view of the physical unfolding of a higher-order singularity where the streamwise pressure gradient and the shear stress gradient, T , vanish simultaneously. The resulting unified picture is given in Figure 3-29 displaying the various properties (bifurcation sets and phase portraits) of the unfolding. Since parametrisation in terms of k and ft appeared to be possible, the unfolding is actually governed by two parameters and has codimension two.

-176-

structurally stable phase portraits

>V- j ">^x^. 1

1 n^l
l_

Fig. 3.29a. Unfolding of Eq. 3.36, ]i. > 0, bifurcation sets and structurally stable phase portraits.

-177-

Fig. 3.29b. Codimension-two degenerate phase portraits of Eq. 3-36.

Cusp point bifurcations

Global bifurcations

Saddle point bifurcations

wrs
I-5 a

W
nr-iib

^2-s:
i-i'

-J\^~

^\c
ffa-Db j,

fV.

V-A>

w
^

V.
a-?b2

Db-?b1

m-m'

A
I'-b1

I'-m

E'-U'
b2-2b1

2 <^->v
a'-Hb'

XI-H

~>r^ ^
I'-H

m'-ib'

Ha'-H

Fig. 3.29c. Codimension-one degenerate phase portraits of Eq. 3-36.

-1788. Viscous flow near a circular cylinder at low Reynolds numbers


8.1. Description of flow topology Although the results presented above have been derived for flat walls, it may be attempted to discuss the steady incompressible flow behind a circular cylinder as an example of bifurcation at zero pressure gradient. We know that the nondimensional quantities describing the flow field will depend only on the the Reynolds number R = p.D.U^/u (where D is the diameter of the cylinder and U undisturbed free stream velocity) and not on D, U p and u separately.

The value of R characterizes the flow field around the cylinder and it is of interest to recall how the flow topology changes with Reynolds number. At small Reynolds number (R < 1) inertia forces are negligible compared to viscous forces over most of the flow field. The dominant process in the flow is the diffusion of vorticity away from the body. At very low Reynolds number (R << 1) the vorticity spreads out fairly evenly in all directions so that the flow has foreand-aft symmetry near the cylinder. This symmetry is reflected very well by Oseen's approximation of this flow.

20

40

60 80 Re=UD/v

5 10

50 100 500 Re=UD/v

(a) length of the wake

(b) position of separation point

Fig. 3-30. Wake geometry of circular cylinder Taneda (1971) Koromilas & Telionis (1980) 1 Homann (1937) theory: Pruppacher, Le Clair & Hamielec (1970)

-179At moderate Reynolds number, specified very roughly as 1 < R < 100, inertia forces and viscous forces become comparable in magnitude and neither of them can be neglected. Vorticity diffused away from the body will be carried downstream and disturbs the afore-mentioned fore-and-aft symmetry, see Batchelor (1970) . Furthermore there appears to be a definite Reynolds number, R = R , above which a region of slowly recirculating fluid occurs, forming a wake immediately behind the cylinder; if R is further increased this wake region becomes longer. Taneda (1971) and Koromilas and Telionis (1980) have measured the length of the circulating region for different Reynolds numbers and their results suggest that this region first appears from R = 5-2 onwards, see Figure 3-30a- The experimental observations compare well with theoretical results as obtained from a numerical solution of the steady state NS-equations by Pruppacher, Le Clair and Hamielec (1970). A similar conclusion may be drawn from Figure 3.30b which displays some numerical and experimental results concerning the location of the separation point on the cylinder surface for various Reynolds numbers. 8.2i_SYmmetrical bifurcations The changing flow topology at a critical value of the Reynolds number will be reconsidered and seen as a local bifurcation occurring at the rear of the cylinder. At the rear a second stagnation point appears and the flows over the upper and lower side meet and leave the surface of the cylinder. Due to symmetry the pressure gradient along the surface will be zero in the rear stagnation point. The flow near this point can be described, at least approximately, by Eq. (3-10) which, for symmetrical flow conditions: u(x,y) = -u(-x.y), v(x,y) = v(-x,y), reduces to x = a_x + a-xy + a_x* - r a_xy* + 0(4) (3.40) y = - a2y - i a5y* - \ a?x y + | y + 0(4) The streamline y = 0 may be regarded as an approximation of the cylinder surface. Separation from this surface occurs at (x.'y) = (0,0) which is a hyperbolic
s

-180saddle p o i n t i f a ? * 0. Due to symmetry the shear s t r e s s and the pressure gradient along the wall will vanish there. For a. = 0 Eq. (3'JO) shows a higherorder s i n g u l a r i t y a t the separation point, indicating a structural i n s t a b i l i t y of the flow pattern probably corresponding to the c r i t i c a l s t a t e R = R . Flow p a t t e r n s o c c u r r i n g a t n e a r - c r i t i c a l Reynolds numbers may be found by unfolding the higher-order singularity (a_ = a_ = - 1 ) :
-xy -

!*
under the restriction of symmetrical flow conditions. This means that uu1 = u, u- = 0 so that the number of b bifurcation parameters is diminished and that the following one-parameter unfolding results

U2x - xy 1 2 p2' v* 3

where u ? denotes small perturbations of a_ with respect to a. = 0, Symmetrical bifurcations as caused by p 2 a r e shown in Figure 3-31a-

rear of cylinder

streamlines

"r
vorticity

Fig- 3-31- Symmetrical bifurcations at the rear of the circular cylinder in steady flow.

-181Obviously the case u_ = 0 reflects a c r i t i c a l state very similar to occurring at R = R . Beyond the c r i t i c a l state (u2 > 0) a cluster of singularities (saddles and centers) arises forming a closed recirculating region behind the cilinder; three stagnation points S.: (0,0), S- _: (Ju_, 0) appear on the surface and three stagnation points appear in the flow field: S^: (0, 4 u o ' 6u2/7), see Fig. 3.31aThe relative positions of the saddle points S2, S, and Su determine the general shape of the recirculating region. In the limit p_ 0 this region shrinks in an 'oval' fashion such that i t s dimension normal to the cylinder (S.) is smaller than along the cylinder surface ( s ) . The ratio of the principal dimensions, /s = Jpp, m a y be expressed in local variables
a n d C

that

2^^?^'

-J-6r .T
_ 5
x x x

io iii \
U-HI)

2p
XX

and will depend on the Reynolds number. Thus unfolding theory leads to the conclusion that /s 0 if R R number such that u- 0 if R + R .
2
c

from post-

critical values and that the bifurcation parameter u_ is related to the Reynolds

Another interesting aspect of the unfolding concerns the vorticity distribution o)(x,y) near the rear part of the cylinder. For symmetric bifurcations (u. = u, = 0) , equation (3.^0) can be used to derive the approximate vorticity distribution. <o(x,y) = -VUx + 2xy + x3 + | xy2 + 0(4) Lines of constant vorticity are displayed in Fig. 3-31b and yield the conclusion that the vorticity pattern near the rear stagnation point is mainly determined by the position of a saddle point on the axis of symmetry at y = 5 P 2 * 0(ul). This saddle point is found in the actual flow domain (wake region) for u_ > 0 only. The level curve of zero vorticity: < o = 0, has a parabolic shape, and intersects the surface of the cylinder in the separation points S . . and S_. The formation of the wake beyond the critical state R = R appearance of a saddle point in the vorticity pattern. concurs with the

-182i2^_AsY5 m etrical_bifurcations i transition scenario's In t h i s paragraph we consider asymmetric b i f u r c a t i o n s of the higher-order singularity:
x = -xy -

x'

y2
y = | -

3
+ |

x2y

in order to investigate some characteristic properties of small asymmetric effects as they can appear if the flow near a circular cilinder is observed experimentally. Actually, deviations in the uniformity of the oncoming flow, imperfections of circular shape and intruding measurement techniques may introduce irregularities disturbing a perfectly symmetric flow pattern. Moreover, at higher Reynolds numbers unsteadiness and flow instability may appear so that symmetrical flow conditions, should they appear, cannot be maintained in time (Von Karman vortex street). The influence of asymmetric disturbances on the steady flow near the critical state may be studied from the influence of the bifurcation parameters u1 and p.. on the physical unfolding:

x = p

+ P

2 X + p 3 y " x y " x'

1 y2 3 u y+ y = - 2 2 3 + 2 x'y Note that the bifurcation parameters u- and u_ may be identified with the shear stress T and the circumferential pressure gradient, respectively, at the rear of the cylinder. The phase portraits and bifurcation sets of this unfolding are extensively studied in paragraph 3-7; a concise assembly of the results is provided in Figure 3-29- In order to apply these results to the flow at the rear of the cylinder we only need the phase portraits in the semiplane y 0. The portraits at y i 0 together with the relevant bifurcation sets are reproduced in Figure 3.32 using the scaled variables k = u_(^-) and i = p,(^-) which are related

to the circumferential gradients of shear stress t and pressure p respectively. It shows a great variety of flow patterns which can occur at the rear of the

-183cylinder and also various possibilities of sequences of flow patterns, when increasing the Reynolds number. In order to discuss the asymmetric bifurcations in a systematic way, we prefer to give therefore more attention to sequences of flow patterns than to the individual flow patterns. However, sequences can be chosen in various,ways. Here we want to discuss those sequences which appear if the bifurcation parameter u_ (related to the circumferential shear stress gradient) increases whilst y. and u_ are held constant. This idea is taken from paragraph 8.2 where a similar approach is followed to discuss the symmetrical bifurcations (p1 = 0, u_ = 0). Moreover, in that case an increase of u ? could be associated with an increase of the Reynolds number.

Fig. 3.32. Asymmetrical bifurcations at the rear of the circular cylinder in steady flow.

-184Consulting Figure 3-32 one observes that a continuous increase of u p , and also of k, causes the intersection of a sequel of bifurcation sets of d i f f e r e n t type (B s , Bc , Bg ) . I t involves the e x i s t e n c e of a s c e n a r i o of s e q u e n t i a l flow p a t t e r n s , linked t o g e t h e r by b i f u r c a t i o n s of s t r u c t u r a l l y u n s t a b l e flow situations. The order in which consecutive b i f u r c a t i o n s are passed through determines a p a r t i c u l a r s c e n a r i o , moreover several scenario's seem to occur. Actually five different scenario's: S. , S 2 . . . . S,- are distinguished in Figure 3.32. All these scenario's reveal the common feature that asymmetric disturbances as introduced by p. and p.. are qualitatively unobtrusive as long as p . < 0, that i s to say i f the c i r c u m f e r e n t i a l shear s t r e s s gradient T remains negative. With respect to the symmetrical situation the flow p a t t e r n i s s l i g h t l y deformed in the sense that the separating streamline leaves the body obliquely dividing the wake flow into an 'obtuse' region and an 'acute' region. If u~ i s increased beyond u_ = 0 the flow behind the cylinder is subjected to a sequence of bifurcations, making the influence of p. and p , more severe. Depending on the order i n which b i f u r c a t i o n s of different kind succeed each other, various scenarios can be observed in Figure 3-32. Let us e l a b o r a t e some d e t a i l s of these scenario's, for example those occurring in S. i f Up increases beyond Up = 0. At pp = 313J (k = 3) a saddle point b i f u r c a t i o n i s passed, announcing the formation of a separation bubble in the 'obtuse' part of the flow. Next a global b i f u r c a t i o n appears which turns the separating streamline counter-clockwise so that the bubble i s now in the 'acute' region. Finally a cusp point b i f u r c a t i o n terminates the sequence and generates a closed flow domain with circulating flow between the separation bubble and the separating streamline. The resulting flow pattern i s topologically different from that in a symmetrical s i t u a t i o n where two c o u n t e r r o t a t i n g v o r t i c e s form a c l o s e d domain with circulating flow behind the cylinder. The occurrence of a free c i r c u l a t i n g region under steady flow conditions i s e s s e n t i a l l y an asymmetric e f f e c t , s i g n a l l i n g vortex shedding t y p i c a l for unsteady quasi-periodic flows at higher Reynolds numbers. Considering the s c e n a r i o ' s S_, S.., Sj. and S- similar conclusions may be drawn from the possible sequences of flow patterns which can occur if p ? i n c r e a s e s to p o s t c r i t i c a l v a l u e s . A u n i f i e d p i c t u r e of these sequences ( t r a n s i t i o n scenario's) i s shown in Figure 3-33-

-185-

transitional " flow patterns

Cusp point bifurcation Saddle point bifurcation Global bifurcation

Fig- 333- Transition patterns in the steady flow of a circular cylinder. On the b a s i s of these r e s u l t s we conclude that asymmetrical effects introduce the p o s s i b i l i t y of more than one t r a n s i t i o n s c e n a r i o . Moreover, such an asymmetrical transition includes several bifurcations suggesting that a certain range of Reynolds numbers has to be passed through before a fully developed wake region i s established behind the cylinder. Lastly, the asymmetrical e f f e c t s include the formation of a free region of c i r c u l a t i n g flow, reminding of an essential feature of vortex shedding, however without including the actual effect and unsteady nature of vortex shedding.

-1869. References
Batchelor, G.K. (1970) An Introduction to fluid dynamics, Cambridge University Press, 255-263. Danberg, J.E. and Fansier, K.S. (1975) Separation-like similarity solutions on two-dimensional moving walls, AIAAJournal, Vol. 13, no. 1; 110-112. Dean, W.R. (1950) Note on the motion of a liquid near a position of separation, Proc. Cambridge Phil. Soc. 46, 293-306. Van Ingen, J.L. (1975) On the calculation of laminar separatrix bubbles in two-dimensional incompressible flow, AGARD Conf. Proc, no. 168, Gttingen. Inoue, D. (198l) MRS criterion for flow separation over moving walls, AIAA-Journal, Vol. 19. no. 9, 1108-1111. Koromilas, C.A. and Telionis, D.P. (1980) Unsteady laminar separation: an experimental study, J. Fluid Mech., Vol. 97Legendre, R. (1955) Decollement laminaire rgulier, Comptes Rendus, Acad. Sci. Paris 241, 732-734. Moore, F.K. (1958) On the separation of the unsteady laminar boundary layers, in Boundary Layer Research (ed. H. Grtler), Springer Verlag Berlin, 296-3IO. Oswatitsch, K. (1957) Die Ablsungsbedingung von Grenzschichten, Symposium on boundary layer research, IUTAM Freiburg. Pruppacher, H.R., Le Clair, B.P. and Hamielec, A.E. (1970) Some relations between drag and flow pattern of viscous flow past a sphere and a cylinder at low and intermediate Reynolds number, J. Fluid Mech. Vol. 44.

-167Rott, N. (1956) Unsteady viscous flow in the vicinity of a stagnation point, Quart. J. Appl. Mech., Vol. 13, 444-451. Sears, W.R. (1956) Some recent developments in airfoil theory, J. Aeronaut. Sci., Vol. 23, 490-499. Sears, W.R. and Telionis, D.P. (1957) Boundary layer separation in unsteady flow, SIAM, J. Appl. Math., Vol. 28, 215235. Takens, F. (1974) Singularities of vector fields, Publ. Math. IHES, 43, 47-100. Taneda, S. (1971) Visualization experiments on unsteady vicous flows around cylinders and plates. In: Recent Research on Unsteady Boundary Layers (ed. E.A. Eichelbrenner) , Vol. 2, Les Presses de 1'Universit Laval, Quebec. Telionis, D.P. and Werle, M.J. (1973) Boundary layer separation from downstream moving boundaries, J. Appl. Mech., Vol. 40, 369-374. Tsahalis, D.T. and Telionis, D.P. (1973) The effect of blowing on laminar separation, J. Appl. Mech., Vol. 40, 1133-H34. Williams, J.C. Ill (1977) Incompressible boundary layer separation, Annual Review of Fluid Mechanics (ed. M. van Dyke), Vol. 9, 113-144.

-189-

Samenvatting
De topologische struktuur van stromingsvelden en de wijze waarop struktuurveranderingen in dergelijke velden kunnen optreden vormen het onderwerp van dit proefschrift. Bij de behandeling van dit onderwerp is veelvuldig gebruik gemaakt van de kwalitatieve theorie van differentiaalvergelijkingen; elementen van deze theorie, zoals singulariteiten en bifurcaties in vectorvelden zijn daarbij frekwent toegepast. De kwalitatieve theorie vindt zijn oorsprong in het werk van Poincar (l880); de theorie is daarna verder ontwikkeld o.a. door Birkhoff (1927, dynamische systemen), Andronov (1937. bifurcaties in tweede-orde systemen), Lyapunov (19^9. stabiliteit) en Arnold (1963. bifurcaties, normaalvormen). In dit proefschrift is het onderwerp beperkt tot stationaire stromingsvelden welke beschreven kunnen worden in een systeem met twee onafhankelijke variabelen; met name worden wrijvingsloze niet-lineaire conische stromingen (supersoon) en onsamendrukbare visceuze stromingen langs vlakke wanden behandeld. De topologische struktuur van een stromingspatroon (bestaande uit baankrommen van het snelheidsveld) wordt in grote mate bepaald door het eventuele optreden van singulariteiten in het patroon. Zo zijn aantal, aard, stabiliteit en onderlinge positie van deze singulariteiten belangrijke struktuurbepalende elementen van het stromingspatroon. De singuliere punten worden op gebruikelijke wijze onderscheiden in elementaire (struktureel stabiele) singulariteiten en in hogere-orde (gedegenereerde) singulariteiten. De hogere-orde singulariteiten zijn struktureel onstabiel in die zin dat kleine verstoringen in het snelheidsveld het karakter van de singulariteit aantasten waardoor een verandering van de topologische struktuur ontstaat. Dit verschijnsel heet bifurcatie, hetgeen duidt op het zich vertakken van een struktureel onstabiele situatie in struktureel stabiele varianten. Stromingsvelden welke door bifurcatie ontstaan kunnen weer struktureel stabiel zijn; ze zijn te beschouwen als lokale oplossingen van de geldende stromingsvergelijkingen. Aldus opgevat vormen de hogere-orde singulariteiten in het snelheidsveld een sleutel tot het genereren van dergelijke struktureel stabiele oplossingen met lokale geldigheid. Het opzetten van een klassifikatie van hogere-orde singulariteiten in stromingsvelden te zamen met bijbehorende bifurcaties ligt daarom voor de hand. Dit proefschrift wil hiertoe als volgt een aanzet geven.

Hoofdstuk I biedt een beknopt overzicht van enkele belangrijke elementen van de kwalitatieve theorie van differentiaalvergelijkingen. Veel gebruikte begrippen zoals fasebeeld, singulier punt, topologische struktuur, strukturele stabiliteit

-190en bifurcatie worden hier aangereikt en toegelicht. Ook de theorie over centrumvariteiten wordt vanwege haar toepassingsmogelijkheden, in dit hoofdstuk kort beschreven. In hoofdstuk II wordt een klassifikatie van stromingspatronen in de buurt van conische stuwpunten in niet-lineaire conische stromingen gegeven. Naast de bekende typen conische stuwpunten zoals knooppunt en zadelpunt, zijn verder gevonden: zadel-knoop, topologisch zadelpunt, topologisch knooppunt en scheef zadelpunt. Met behulp van deze conische stuwpunten worden verschillende stromingsfacetten in conische stromingsvelden verklaard en toegelicht zoals: het 'lift-off' verschijnsel bij cirkel kegels onder invalshoek, het genereren van conische stuwpunten, 'niet-visceuze' loslating van de stroming langs kegeloppervlakken en het conisch stroomlijnenpatroon nabij het hoekpunt van een uitwendige hoek of deltavleugel met geknikt oppervlak.

In hoofdstuk III wordt een beschouwing gegeven over de topologie van een stationaire stroming van een visceus medium in de nabijheid van een wand. Lokale oplossingen van de Navier-Stokes vergelijkingen worden geconstrueerd met behulp van de kwalitatieve theorie. De bijbehorende stromingspatronen zijn verkregen als bifurcaties van gedegenereerde hogere-orde singulariteiten. Naarmate de degeneratiegraad van de singulariteit hoger is blijken stromingspatronen te ontstaan waarvan de topologische struktuur komplexer wordt. Vervolgens wordt in hoofdstuk III een klassifikatie gegeven van stromingspatronen, die in de buurt van een stilstaande of bewegende wand kunnen optreden; deze klassifikatie biedt een verscheidenheid aan mogelijke stromingsvormen. Naast bekende patronen, zoals de klassieke Oswatitsch-Legendre oplossing voor loslating, doen zich nieuwe strukturen voor; o.a. die welke geschikt blijken voor de beschrijving van: - het ontstaan van lokale loslaatbellen in uniform geslaagde stromingen; - de interferentie van een loslaatbel met een stroomafwaarts gelegen, tegen de stroom in bewegende, secundaire loslating; - loslatingsverschijnselen in de stationaire stroming langs een bewegende wand; - de strukturele stabiliteit van de stroming in een 'zadelpunt-driehoek'; - de vorming van asymmetrische stationaire 'eddies' in het zog van een stomp lichaam. De resultaten van het onderzoek als beschreven in dit proefschrift ondersteunen de stelling dat in stromingsvelden optredende, hogere-orde singulariteiten, hoewel nauwelijks waarneembaar in de praktijk, belangrijke bouwstenen zijn voor het konstrueren van lokale oplossingen van de geldende stromingsvergelijkingen. Als zodanig kunnen ze bijdragen tot het verschaffen van inzicht in de fysica van stromingsverschijnselen.

-191-

About the author


Pieter Gerrit Bakker was born May 13th, 19^5. in Groningen, The Netherlands. In 1966 he obtained the diploma of aeronautical engineering of the HTS (College of Engineering) in Haarlem. In September 1970 he joined the Department of Aeronautical Engineering of the Delft University of Technology. As a member of the technical staff of the Laboratory of High Speed Aerodynamics, he was involved with lecture demonstrations and experimental research in supersonic aerodynamics. In February 197^ he graduated cum laude as 'ingenieur' in Aeronautical Engineering at the Delft University of Technology, on a thesis concerning nonlinear conical flows in gasdynamics. Since March 197^ he is a member of the academic staff of the Laboratory of High Speed Aerodynamics, with a main task in education and research in transonic and supersonic aerodynamics.

You might also like