1. Fundamentals in Continuum Mecha Certain concepts and definitions are basic to the study of continuum mechanics, and they should be thoroughly understood at the outset. Below, a description of the motion of flow is given in three dimensional space with respect to the reference frame or relative to the rotational frame. Dynamics involves the frequent use of derivatives of scalars, vectors and tensors. Ve locity and acceleration are the time derivatives, which are important kine matic parameters necessary to set an equation of motion in Newtonian m chanies, on which continuum mechanics is based. Also given are forces connected with space derivatives in regard to displacement gradient and relative strain, which are other important aspects of continuum mechanics. In this chapter, definitions of stress tensors and strain tensors are also pro- vided, and will be developed in more detail in the chapters to follow. 1.1 Dynamics of Fluid Motion When we deal with fluid motion, in many fluid engineering cases the dy- namics of a molecule, or the molecular structure ofthe fluid body, does not explicitly come into effect. At the scale of molecular motion, propert the fluid body, such as density, are typically subject to extreme varia with respect to the instantaneous distance of the frame. While, for the mo- tion of fluid flow, the macro-motion with the scale of flow channel or e ternal object takes place, thus we may apply the “continuum hypothesis”, with which the fluid body has a continuous structure in the instantaneous frame of space, as schematically indicated in Fig. 1.1. In Fig. 1.1, let us denote Ln as the small scale (molecular scale), which can be taken as the mean free path of the molecule; Z, as the large scale, which can be the characteristic length of the geometric cont ‘over, there may exist an intermediate length scale L,, where a certain fect of a molecule or the molecular structure retains the properties of fluid, In order to quantify the effect of the scale in the properties of fluid, and consequently to the dynamics of fluid motion, we will take the ratio be- tween the actual characteristic length of flow geometry, typically L,, and suration of fluid motion. Mor6 1 Fundamentals in Continuum Mechanies the mean fr cules) Z,,, such that we get the following formula where I is called the Knudsen number. e path of the molecules (the correlation length of the mol “Continuum Mechanics” in a general sei particular, is normally valid when TP << 1, where the continuum hypothe- sis can be assured. Henceforth we shall make two assumptions: first, that in every case, the flow of fluid has a small Knudsen Number, with which the scale of momentum of flow is far longer than the correlation length of the molecules; and second, that the fluid body has a continuous structure. sul 5 Proper Molecular seale | Intermediate seale | Large seale SSL Ly 4, L Fig. 1.1 Property variation with scale (as typically seen with properties such as density p) It is interesting to mention, although we will not deal with the problem in this text, that there is a field of study that deals with small scaled flow phenomena in continuum mechanics. These phenomena have been dubbed fluid flows in a micro-channel, or micro-fluid-mechanics as it is more commonly referred to. The reader may wish to refer to a more detailed d scription provided by Kim and Kavila, 1991, and Tabeling, 2005 ‘The motion of a fluid can be perfectly determined, when the velocity at ‘every point of the space is occupied by fluid motion. Therefore, to express the velocity with independent variables, there are two distinct methods, the so-called Eulerian and Lagrangian specifications1.1 Dynamics of Fluid Motion 7 One method is to trace the motion of a particle in space with time. In particular, a particle in fluid is called “fluid particle”, which is a subdivi- sion of the fluid around a specific point xx. This method is called Lagran- gian specification. The position vector of the fluid particle, considering a motion relative to a given frame of reference at time 1, can be expressed x= x(xy.0) q2y where xo and f are independent parameters, and xy is the original posi- tion at f= 0, xo =x(x0,0). The velocity w and acceleration a at time 1 can be written similarly as ax wale =e (13) ou a= als) (14) ‘As seen in Fig. 1.2(a), Lagrangian specification describes the motion of a body of mass (fluid particle) and its variation of flow state along the parti- cle path, ‘Another method is to give the spatial distribution of flow sta function of spatial coordinat x,y,2)ia,y,Zare Euler variables) and time 1, wher ependent variables. This method is the variation of flow state in a position x (position vector in spatial coordinates x ) at a given time? , not describing the behavior of each partic! ig. 1.2(b). In the Eulerian specification the velocity w can be expressed as u=ucx, 1) as) ‘The expression of the acceleration @ in an Eulerian specification is given by differentiation following the motion of a fluid particle and the rate of change of the velocity of that particle with respect to time uxt Ax, f+ AN—u(x, 1) _ Ou Ax fim BOE AN Oe A SH tim SE ar ar tim ) (1.1.6) Particularly in fluid mechanics, @ in Eq. (1.1.6) is called the material de- ind is often expressed as8 1 Fundamentals in Continuum Mechanics , Sauy \u aan la bu Dr In general, the rate of change over time of material parameter A, where 4 can be a scalar, vector or tensor associated with the (material) point at time 4, can be expressed by the material derivative (the differential operator D/Dt is also called Lagrangian derivative) as Pabst) 2 Alx.t)+ (u-V)Alx.t) (18) (@) Lagrangian specification (») Bulerian specification a : Particle Flow field (e1) position vector x Fig. 1.2 Description of fluid motion ‘The derivative defined in Eq, (1.1.7) can be further written, using vector idemtties (see section of Appendix B-5), as follows Du a 119) a (1.9) ‘The term (w-V)w or (w-V)A, operating in Eqs. (1.7) and (1.1.8) respee- tively, are called the convective term, which express the fact that, for time independent flow @/@t =0,, the fluid properties, such w or A , depend only upon the spatial coordinates x. Namely, the changes in fluid properties are due to the changing spatial position of a given fluid particle as it flows. The terms Gu/ér, 24/61 are the Eulerian time derivatives evaluated at a position x.1.1 Dynamics of Fluid Motion 9 Having established the conceptual and mathematical representation of the kinematic description of motion, we will now consider the decomposi- tion of motion. This will be done in order to establish the concept and mathematical description of that rate, An important concept in a motion of continuum is to know how the deformation can be expressed from the ve- locity field w= u(x,t). Fig. 1.3 Material line L change Let us consider the configuration of an infinitesimal element L(t) of a straight material line, It undergoes translation, rotation, and stretching, as a result of the nonuniform velocity field of w= dx/dr . As time elapses, the material line element Z(t) moves to the new positions P and Q, where L(¢) becomes L(¢+ di). From the figure in Fig. 1.3, itis clear that L(t + dr) = [fx(s) + L(n)} + u(x + Lotte] [x (0) + a(x tar] = L()+u(x-+L,t}tr —ulx,t)de (110) so that, the changing rate of L(t) as ¢ 4+ df can be written as wes 10) af tt) alee) = 2 Gay where the right hand side of Eq, (1.1.11) is a Taylor series expansion for u(x,t) around £(). Equation (1.1.11) can be further modified as10 1 Fundamentals in Continuum Mechanics (1.1.12) Ax, since we are taking L(¢) as an infinitesimal line element, so that we have Hc+a)=u6)_ax/, Te al (1.13) and Au(_ au _.,\ m MA = vy 1.1.14) aL) dina oae } any E js called the displacement gradient tensor and Vu is called the velocity gradient tensor. Using the notation defined in Eqs. (1.1.13) and (1.1.14). Thus, the change rate of material line element L(t), which is «given by Eq, (1.1.11), will be written by the formula AxV. {1 yale, 7 ax {Warve ) $ave } =Ax-e+0) (1.1.15) where superscript T denotes the transpose of tensor. It is noted here that in ‘expanding Eq. (1.1.15) the velocity gradient tensor Vu, arbitrary second ‘order tensor is decomposed into symmetric and skew-symmetric (or anti- symmetric) parts. We now define the rate of strain (or the rate of deforma- tion) tensor, as the symmetric part of the velocity gradient tensor e=Lousva!) (16) and the vorticity (or spin) tensor, as the skew-symmetric part of the veloc- ity gradient tensor (Vu-Vu") (1.17) ‘Thus, the change rate of L(t) due to translation u(x,t) can be straining Ax-e and rotation Ax-@. Consequently the velocity gradient tensor can bbe written by1.1 Dynamics of Fluid Motion 11 Vuze +o (11.18) sors € and @ in contin- st place. Referring Eq, (1.1.16), € can be written by the Cartesian suffix convention for three val- ues, = 1, 2 and 3, or j= 1, 2 and 3, which correspond to x, y and 2 In order to gain more physical insight for t uum mechanies, we will examine e in the respectively as follows (1.19) where 6,8, is the unit dyad (see Appendix B-1), and the components of the rate of strain tensor e, is presented by AG) ERP le oe & AGer) lees) tis ly known from Eq, (1.1.20) that ¢, is the symmetric and has 6 in- dependent components, ic. e1,e32 ,e8,@2 = en ,en =en andes: =x. The orthogonal components e11, e2 , and ex: are the rate of elongational strain, due to a local deformation of fluid in stretching or contraction in x,y, and 2 axis respectively. In addition the off-orthogonal components es, ex, and ex: (as well as ex1, 2, and es) are, on the other hand, due to a local deformation of fluid in shearing in the plane of x-y, y—z, and 2—x respectively. Figure 1.4 gives an idea of the deformation rate occurring in fluid in x—y plane. Figure 1.4 (1)-a shows the simple elongational (or ex- tensional) flow field as one of typical stretching and contraction of flow, where the rate of elongational strain appears. Furthermore, in order to make the point clear, in Fig. 1.4 (1)-b the stretching of a fluid element in the elongational flow field is indicated. Similarly in Fig. 1.4 (2)-a the dia- ‘gram shows the simple shear flow field, where the shear strain takes place and (2)-b a sketch of the shear field is displayed, where a fluid element is sheared in the flow direction,12 1 Fundamentals in Continuum Mechanics. ‘The symmetry of the tensor e guarantees that there will always be three mutual orthogonal orientations of = @X/ér, for each of which the corresponding rate of deformation is either elongation or contraction, that is the principal rates of deformation, to which we can a coordinate transformation, which give the proportional rate of increase of an infinitesimal material volume ign ef =e, wariant for and een. The summations of ef, e$ and ef are Vewadivanel +e5 +e, (2 (1)-a Simple elongational (2)-a Simple shear flow field (or extensional ) flow field (@)-b yb ol Fig. 1.4 Local deformation of fluid This quantity defined in Eq. (1.1.21) is called the divergence of the veloc ity field of u(x,t). Furthermore, a definition that is independent from a coordinates system can be given by1.1 Dynamies of Fluid Motion 13 1 Dor) Di diva = o [paar aa Equation (1.1.22) is called the Euler’s relation, which implies the physical rate of change over time of the volume of moving fluid particles per unit volume. It is noted that to write the volume integral j v-udV from the surface integral ja udS the Gauss’ divergence theorem is applied. ‘The skew-symmetric part @ given by Eq. (1.1.17) of the velocity gra- dient tensor Va can be similarly written by the Cartesian suffix conven- tion as (1.1.23) ° su, (& 1) (eu a) Flan ee (0am) (an én a) Ge fo a om] =s|on 0 on “lou on 0 1.24) Components, which appear in Eg. (1.1.23), as the components of @=V xu, are related with the following formula vane 28 )n2(% = bay + bx0 +8505 (1.1.25)14 1 Fundamentals in Continuum Mechanics It may be useful to consider «= V xu in a little more detail for the sake of coupling equations in the following chapters. The nature of Va can be examined by introducing the altemator or alternating unit tensor £28,226, as follows Va ene 5 ou, g le Oe wend Es cubs Es ends ax; x; a (2.1.26) ‘Thus, in comparison with (1.1.23) and (1.1.25), @ vector «(in this Vx) with components @, @; and , can be reduced from a gei Gu,féx,, Oufax, Yu or grad w), whose components of skew-symmetric part (in this case @) are written by second order tensor $ (in the present 0-0 a 0 =a 2 -o % 0 Note that the vector @ =(«,a,0) is called the pscudovector of the tensor S or simply the veetor of tensor $ . See Exercise 1.4 By employing the alternator e, the vector @ may be found from S, more specifically from the skew-symmetric part S, of S, denoting S.=0/2 while the symmetric part $, of S may be expressed by o/2.. The following relationships articularly useful; (1.1.28) Si=seo (1.1.29) Note that from Eqs. (1.1.28) and (1.1.29) the pseudovector « is zero, , ifa second order tensor $ is symmetric S$. =0, and vice versa.1.1 Dynamies of Fluid Motion 15, For the sake of clarity and convenience, the unit tensor I and the al- temator £ are defined below. I is defined such that 6,5, (1.1.30) it it 131) ¢ is similarly defined by £=88 2:0 (1.1.32) where 6, is the Eddington notation and é,8,2, is the unit polyadic. Thus, the alternator is a polyadic. ¢;¢ is defined in the following manner 1 if jk =123,231or312 ou 4-1 if fk =321,132,0r213, (0.1.33) 0 if or j=k Its useful for e to be alternatively expressed with the form £=(6:6:6, +6,6,6, +866, -86,8, -1 6,) (1.1.34) After giving mathematical formalities for the tensor o , we will see how @ may cause a local deformation in the flow field u(x,t). The de- formation due to @ can be examined from Eq. (1.1.11) as = Le =a) Li =|L-{ 4) ar 1.1.35) AX =L(t+dt) >| (&) (1.1.35) Note that (@u/2L), is a skew-symmetric part of the velocity gradient ten- sor. It is further noted that by taking an infinitesimal time duration dt ~ & imal line element L=Ax = 6x, we can obtain the deforma- and on i tion in x-y plane, for example 0 -0, 0 en) 0 0 0 (1.1.36)16 1 Fundamentals in Continuum Mechanics. ‘The implication of Eq, (1.1.36) is, in the x—y plane, that as indicated in Fig. 1.5, the fluid element é& 6 is rotated around the = axis with its an- gular velocity Q, = 4/2. Thus, as easily speculated from Fig. 1.5, the vorticity tensor represents a rigid body rotation of fluid element with angular velocity © = «/2.. The pseudovector of the velocity gradient ten- sor, ie. @=Vxu=—e: Var, is called the vorticity vector. The vorticity vector q is an important flow parameter in fluid met Fig. 1.5 Rigid body rotation of fluid element é—6 1.2 Dynamics in Rotating Reference Frame In consideration of kinematics let us explore the relationship between an inertial and a rotating reference frame. For brevity, let the rotating r cence frame be rotated with a constant angular velocity with respect to the inertial reference frame, supposing no translation of the rotating refer- cence frame to the inertial reference frame, see Fig. 1.6. It is noted that we adopt a right-handed orthogonal coordinates system, where Q>0 is an angular velocity vector, which rotates to the direction of a right-handed screw. The position vectors x, and x, of a material point x in the iner- tial frame and rotating frame respectively are related as X a2 where Q is a rotation tensor, which is an orthogonal tensor with a relation of QT Q=H, and Q?=Q* (where Q is an unitary matrix); Q-!1.2 Dynamics in Rotating Reference Frame 17 denotes the inverse of the tensor Q and Q” denotes the transpose of the tensor Q Material point Fig. 1.6 Rotating reference frame with inertial reference frame In order to correlate two frames kinematically, we shall consider a ve- locity and acceleration at the material point. To start with, apply D/Dr , the material derivative, to Eq, (1.2.1) to obtain the relative velocity u, Dxy +Q Dx. Deo (1.2.2) QT Equation (1.2.2) can be further written with the velocity a, in the inertial frame, letting Dx, /Dé=u, by Axx, Quy (1.2.3) Here we used tensor calculus (for relative position vector x, see Problem 1-1)as DQ gry, PQ ger, Axx, =- 23.97 x, =-28.QHhx, (24)18 1 Fundamentals in Continuum Mechanics It is noted that 2 is the pseudovector of the skew-symmetric tensor (DQ/D1)-Q* in the same manner explained for the vorticity vector @ previously. Next, by taking the material derivative again to Eq, (1.2.3), we ccan obtain the relative acceleration a, of the material point x as follows Du. Dr 0 +22 Qn, 22.94. (axx,)=Q-a; Oxu, + EQ (Oxx/)=Qav (125) Consequently from Eq. (1.1.4), we can obtain the acceleration vector ao, i.e. the acceleration of the inertial frame, by relating Eq. (1.2.5) in the ro- tating frame as ay =a, + 2x u, +Qx(Axx,) (1.2.6) where it is derived when an inertial frame is instantaneously coincident with the rotating frame , as Q=1.. In Eq. (1.2.6), the first term in the right hand side of the equation is the rectilinear acceleration, the second is the Corioli’s acceleration, and the third is the centripetal acceleration. With identical vectors, it is easy to show that the third term of Eq, (1.2.6), ie. the centripetal acceleration can be reduced to the potential form as ax(axx, leep after) aan so that, defining a potential function g , and setting g=Q7r7/2 , where r simply denotes distance from the axis of rotation, the centripetal accelera~ tion can be express Qx(Qxx,)=-VE (1.2.8) It may be worthwhile to note that the operation to reduce the relation (1.2.6) can be simply understood by a vector algebra for a fluid particle ro- tating at a constant angular velocity Q along an axis of rotation in the in- crtial reference frame at a material point x “=, 129) and ty =u, +OXx, (0.2.10) Equation (1.2.9) refers to the instantaneous moment, when Q=1, and Eq. (1.2.10) implies the relative velocity of the fluid particle rotating to the1.8 Material Objectivity and Convective Derivatives 19 inertial frame. Thus, taking the time derivative to Eq, (1.2.10), as we have similarly done to Eq. (1.2.3), we can obtain the acceleration of the fluid particle relative to the inertial frame as ay = De 205, £Ox(Ox Di dt . 2.4 which is exactly the same form as Eq. (1.2.6). The relation given by Eq, (1.2.11) can again be written, using the potential ¢ ay =a, $2, VG 0.2.12) where $=97r7/2 1.3 Material Objectivity and Convective Derivatives (On the microstructure level material elements may be affected by strong electromagnetic field or strong inertial forces; however, on the continuum level the physical characteristic of a material, such as demonstrated by the Hooke’s law (the relationship between the extension and the force can be regarded as a physical property of spring itself), is independent of the mo- tion of the observer. This concept is called “the material objectivity” or “the principle of frame invariance”. Particularly in dealing with the rela- tionship between a deformation and a stress in continuum, so-called consti- tutive equation, this concept is of some importan In order to satisfy the principle of frame invariance, the following lin- ar transformation by E (see Q and E in Eqs. (1.2.1) and (1.3.4) for equivalence) must be satisfied for any arbitrary vectors (say a velocity vee- tor u(x,f)) and second order tensor (say a stress tensor T(x,f)) in the Cartesian reference frame. This can be set in the inertial reference frame in such a way that = Qu to w= Ew! 3.1) 1, TQ to =E-TET (32) where the suffix r denotes the rotating reference frame. Note that scalar properties (such temperature, density, etc.) are alwa Now we will direct our attention to how the time derivative of vectors and tensors are affected in order for the principle of frame invariance to be satisfied. We will consider this problem with respect to the transformation20 1 Fundamentals in Continuum Mechanics ‘of a material line element as displayed in Fig. 1.3, where the material line element L(t)=Ax (in the material reference frame) is transformed to L(t+dr) (in the new space frame), noting that L(t +d) - L(t)=AX and defining the displacement gradient tensors as fim SS (133) so that Eq, (1.3.3) can be further written as AX = E-Ax 3.4) and Axe El AN (3.5) Eventually E is a linear transformation tensor, and in order to verify the principle of frame invariance we will transform a tensor T' (in the mate- rial reference frame) to T (in the new space frame). This transformation can be written T=E.T-E" (1.3.6) and the reverse transformation is defined by Tek. T(E4) 3.7) ‘Thus, the material derivative of T’ in Eq, (1.3.7) will be bY’ PE eine) ee hey se ey! or (ey (ery “ey (13.8) denoting the material derivative, ic. DT'/D1 =". Equation (1.3.8) can be further reduced to the form Ts ET (EY EA (Va T) (EF E(u") (2 = E+. (T=Vu-T-T-Vu) (EA) 13.9) It is noted that to derive Eq. (1.3.9) we have used the relations =-E-E-E*! and Va=éu/éx=E-E". The relations are obtained from Eq, (1.3.4) for the velocity w as (13.10) and1.8 Material Objectivity and Convective Derivatives 21 eu _ yp ex 41 Vua Bak SEE (aan Consequently, the tensor (‘T-Vu-T-T-V.u" ) is linearly transformed by a material line element of E~' , and we denote Eq. (1.3.9) as pee hey (13.12) V is called the upper convective derivative, which is defined as fol- poet vat teva? Dr = ievu-T-Tu" (3.13) In the upper convective derivative, the base vectors are “‘contravariant” ‘base vectors. That is, the base coordinate vectors are parallel to material lines, which are deformed (stretched and rotated) with a material line. In a similar manner, with “covariant” base vectors, that are normal to material planes, where in a deformation each base vector rotates to remain normal and stretches so that its length remains proportional to the area of the material plane to which it is normal, we have the lower convective de- rivative defined by Tati Vu" Ty T-Vu (3.14) If we extend further, we will see that when material lines are in a de- formation with rotational coordinates, we can define the derivative in the following manner T=t-oT+To (13.15a) or alternatively (1.3.15b) where o is the spin tensor and the derivative TT is called the corotational derivative (a) or the Jaumann derivative (b). The corotational (Jaumann) derivative can be gotten from the upper convective derivative directly by setting Vu =e+@ =0+0, and using @ =- 0",22 1 Fundamentals in Continuum Mechanics ‘The three derivatives, Eqs. (1.3.13), (1.3.14) and (1.3.15), ean satisfy the principle of frame invariance, and this can be easily demonstrated to meet the condition in Eq, (1.3.2) for the rotation tensor Q to give DU yep toni BE vata (vu') =2070)-Qvu-9'+9)-0T0" = Q-7-Q7-(Q-VH-Q'42") bt 2022 vet Tvu).Q" OD re (13:16) So, T meets the sufficient condition for the principle of frame invariance. Observe that in deriving Eq, (1.3.16), the following relation was used 2=Q-Q' =-QQ' =-2" (13.17) since we have Q.Q"=, where Q is the unitary matrix, which is written by QQ" =- QQ". It is further noted that the velocity gradi Vir itself does not satisfies the principle of frame invariance as itis that Vu'=Q-Vu-Q7+Q-Q" =Q.V0-Q42, (13.18) 1.4 Displacement Gradient and Relative Strain ‘The displacement gradient tensor E , defined in Eqs. (1.1.13) and (1.3.3), is not generally symmetric and contains both deformation and rotation of a material line. Thus, E itself is not a quantity of the frame invariance and ‘one may have to exclude the effect of rotation of a material line, particu- larly when a constitutive equation is considered. In order to define finite strain tensors, which are free from rotation, we can simply take the square of the length of Ax or AX , for the material lines before and after deforma- tion respectively (see Fig. 1.3), such that ds? =Ax" «A= (E°- AX) -(E-Ax)= AX" (EE ax (14.1) dS? =AN? - AX =(E-Ax)" -(E-Ax) = dx" (E7-E)-de. (1.42)1.5 Reynolds’Transport Theorem 23 Equations (1.4.1) and (1.4.2) contain metric tensors, (E-E7)“! and (ET-E) respectively. Hereafter we define strain tensors, one of which is called the Cauchy strain tensor C defined by C=ECE (143) and the other is called the Finger strain tensor defined by Ct SEE (44) ‘These two tensors are both positive symmetric, and describe the deforma- tion from f to £+At of a material line, which are free from rotation, Con- sidering the character of two tensors, € and C', we can define two closely related relative strain tensors as follows, hate (4s) (1.46) Ye ‘These two relative strain tensors are very useful and are often used in de- riving integral constitutive equations for viscoelastic fluids. 1.5 Reynolds’ Transport Theorem In deriving conservation equations of flow, it is particularly important to consider the volume integral J = j oF (NdV, of which material deriva- tive is defined as DI/Dt. V = V(t) is a closed volume of fluid particles, or otherwise known as a material volume (element) consisting of a represen- tative material line, and F(x,1) is any scalar, vector, or tensor function, Reynolds’ transport theorem concerns the tegral, ie. DI/Dt Before proceeding further, it may be useful to consider the change of a material volume dV from coordinates § at time fo to coordinates x at time 1, where ¢ is the material coordinates, and they are Cartesian coor- dinates, & =(£1,£2,g:). Also let the volume element dV be d&i,dé,dgs of an elementary parallelepiped, as sketched in Fig. 1.7. Due to the fluid motion, this parallelepiped d/o is moved to some neighborhood of the Cartesian point x= .x(G,1) at time ¢, with the volume element of d¥, of change of any volume in-24 1 Fundamentals in Continuum Mechanics whose sides are dn, drz and dry, ie, dV = dridrades. The change of co- ordinates x-=(n,x2,21) must be given to the corresponding coordinates §=(6.8.5) by Ham(Ge8) as.) The sides dry, dry and dr, of the volume element can be given by chain rule as (52) as tn) Fig. 1.7 Change of material volume at time / ‘The resultant volume clement dV” can be calculated by the box product [ ] of vectors’ dx,, dx), and dx, three sides, representing material line elements of dV a [dx dvd, ]= de, -(d, xdv,) = Jdg de, dé, (15.3) where J is called the Jacobian of the transformation of the variables, and is defined as1.5 Reynolds’ Transport Theorem — 25 alvmim) | aa as len an DEES) “OE 0 a ob ea lam am a o& From the relation given by Eq. (1.5.3), J is the ratio of a material vol- ume element to its initial volume as av a, (ss) and dV = JdVo. This is called the dilatation. In consideration of the dilatation, the time derivative (the material de- tvaive) of the volume integral 1= [ F(x,” can be writen (by means of the Lagrangian description) F(x,t)dV = SIF Lege) ar (15.6) Using the definition of the divergence in Eg, (1.1.22), we ean obtain the dilatation’s relative rate of change along a path line of a fluid particle as follows 1 Dier,)_ 1 by “Ta Dr Tr asa Thus, from Eqs. (1.5.6) and (1.5.7) we can obtain the following relation- ship oe tr F@-W fear = f, (eve ahr (53) Did, 1 ‘The formula (1.5.8) is called the Reynolds’ transport theorem, and can be further extended into a number of different forms, using the definition of the material derivative given by Eq, (1.1.7) as26 1 Fundamentals in Continuum Mechanics DI er ae v -(Fu) hav Dt fee ( } (re ar “ =( ars a-ruas Fr f, (1.5.10) It should be noted that the Gauss’ divergence theorem was applied in order to write the surface integral f'~ dS from the volume integral f, ~ dV. ‘The physical picture of the Reynolds’ transport theorem is that the rate of change of the integral of F in Lagrangian description is the sum of the integral of the rate of change at a point, and the net flow of F over the control volume surface in Eulerian description 1.6 Forces on Volume Element ‘There are two kinds of forces acting on a volume element of @ continuum medium. The volume clement taken in a flow field is called the control volume in Eulerian description and equivalently called the fluid particle in Lagrangian description. In both cases, as depicted in Fig. 1.8, “Body forces” as one of the two kinds, can be regarded as reaching the medium and acting over the entire volume. Body forces, which are represented by a symbol g, are due to long-range forces, such as gravitation (with the gravitational acceleration g) or electromagnetic forces, ete. They are usu- ally independent from a deformation of the volume element and are caused by an external field of source. “Surface forces”, of another kind, are to be regarded as acting upon the surface of each part of the volume element. The origins of surface forces are chiefly due to two short-range forces, viscous and elastic forces, those have strong dependence on a deformation of the volume element. The sur- face forces have molecular origin in the vicinity of the surface, and act on internal forces through the surface. Surface forces may also be generated by an externally applied field, such as electromagnetic field, through sur- face coupling. We shall see detailed descriptions of body forces and sur- face forces in later chapters, such as Chapter 7 and Chapter 8. However, at present we will treat the surface forces, with reference to stresses, that can be represented by a stress tensor. We follow to define a general stress ten- sor ina continuum medium, in this chapter, through Cauchy’s fundamental theorem for stress.1.6 Forces on Volume Element 27 Stress is a vector quantity defined as a force per unit area. Let AF bea force exerted across an area dS , on which the unit outward normal vector is acting by the material. From Cauchy’s theorem, the stress v by (16.1) Fig. 1.9 Stresses on surface elements28 1 Fundamentals in Continuum Mechanics In order to give an insight to the nature of the stress fj) at point a, we consider an elementary tetrahedron of the body as schematically displayed in Fig. 1.9, where the front surface dS has a normal unit vector of f. AS elucidated in Fig. 1.9, the small tetrahedron has three of its fac the surface elements are dS; = figydSi, dS2=figjdS2 and ds = ., where figydSs denoting fj) the normal unit vector and dS; the area of each element. We may further indicate the stress vectors over each of three faces, by 1) for 1, 2, 3 and n. Since the surface of the tetrahedron is closed, owing that four faces (=1, 2, 3 and m bind the tetrahedron, we apply the principle of local ‘equilibrium to the stress forces, subjecting to infinitesimal tetrahedron, so that tod S — (td, + #429452 + t)d5s)=0 (1.6.2) ‘Then, in the limit of dS; +0, we can write the components of the surface ‘element in such a way that (AS, .f404 sine dS = dS) =) becomes 10) 89s) (yy tia) otiy JAS (1.6.3) ISs can certainly be true. Thus Eq. (1.6.2) Hj = Fit) + fieyta + Foy)) (1.6.4) ‘The expression in the parenthesis is a dyadic and each term in the paren- thesis has three components, that is Ty Tx Ty faye) =| Tis Be), foe) =| Ta P:8,, faoyoy =| Te [ee (1.6.5) Tis Ts Ts where ff) shows the vector components of 4) on the surface of AyydSp) facing fi) direction, and so on. Using the dyadic notation, Eq, (1.6.5) can be written as Tis Tia Ta T=|Tu Ta Tr 2) Tu Ta Ts =T188, (16.6)1.6 Forces on Volume Element 29 where Tis called the stress tensor, which has components of a second or- der matrix. Therefore, Eq. (1.6.4) can be written altematively with the stress T by tenk, (16.7) (1.68) The expression of Eq. (1.6.6) is called Cauchy’s stress formula, It can be verified by some tensor calculus that #-T=17.A, where TT is the transpose of T. It is, however, only true that when the tensor Tis sym- metric, the relationship of f-T=1"-f can be held. The diagonal components Ti, Tzx and Tis of the stress tensor T are called the normal stresses and the off-diagonal components Tir , Ta Ta, Tis ,Txand Tx are called the shear stresses. When continuum medium is at rest, implying that fluid velocity is identically zero at any given time, any stress acting upon a volume element is called hydrostatic stress, except for very specific Newtonian fluids or electromagnetic medium, which will be introduced in the later chapters. The hydrostatic stress is a normal stress, which is inde- pendent of the orientation. The hydrostatic stress can be expressed by de- noting — p as nT, =-pn, (1.6.9) and this expression yields the following relationship as T, =—P6, (1.6.10) or alternatively T=-pl (6.11) where I is the unit tensor. p in Eqs. (1.6.10) and (1.6.11) can be identi- fied as the thermodynamic pressure in a compressive fluid under assump- tion that the fluid is in equilibrium even when the fluid is in motion. How- ever, in incompressible limit, p can be treated as an independent dynamic variable, retaining pas a pressure. Including the hydrostatic stress, the stress tensor may be written as T, =-pd, +1, (1.6.12)30 1 Fundamentals in Continuum Mechanies or alternatively T=-pl+t (1.6.13) The stress tensor +t, which is often termed as deviatoric stress tensor, may include various contributions, depending upon the physical character of the continuum medium, such as compressibility, viscoelastic nature, and external (such as electromagnetic) field effects, likewise for p mentioned above. A general expression of the total stress T may be expressed by 1, =-p'b, +1, (1.6.14) where p’ and rj are the extended pressure and stress tensor respectively. The mean of Ty is de Im yh =P Sty (1.6.15) and the deviatoric stress zy is defined as ty =T) ~ Tad (1.6.16) In viscous, incompressible Newtonian fluid, ic r) =0 and p* = p, the mean stress is equal to the pressure p as Ta =—p (16.17) ‘This fluid is sometimes called the perfect fluid, Exercise Exercise 1.1 Dyadic Product Vu ‘Vu is called the gradient of vector w and is sometimes written du/x or gradu. Vu is the second order tensor in the Cartesian coordinates system. Show Vu as the dyadic product, using suffix notation of tensor with unit dyads 8, Ans. oExercise 31 Equation (1) shows that Var is a second order tensor whose ij compo- nents are @u, /@u, Exercise 1.2 Convective Term In Eq, (1.1.7), the term (u.V)u=1- Vir is called the convective term in fluid mechanics. Using the vector identities in Appendix B- (B.5-6), re- duce Eq, (1.1.9) Vlu-v) =m. Yo vy Yury ux(Vxv) 9x (Vu) a Ans, Set v =u , which gives u-vu=ty) ~ux(Vxu) Q When the vorticity vector w =V xu , Eq. (2) becomes “ vu Jae @) If the velocity field w is irrotational, ic. @=Vxu=0, u has a scalar po- tential g such that Vo a) and with the scalar potential the convective term will be written ) * va=v(Aver ) © wer) Exercise 1.3 Euler's Relation Proof the Euler Relation given by Eq. (1.1.22) Ds So vew 1 a a wher82 1 Fundamentals in Continuum Mechanics ae, an on 2] ° 8g, 86, O64 86; ~ Ans. We firstly write the following relation, for the velocity gradient @) Bree, ae, D2, With this relation, we are able to set an expression of the time change of the Jacobian as follows Di, Bb a a Di" ag, ag, Ob" 8s, o While Eq. (3) may be formulated by chain rule 28 am _ ay oy oExercise Exercise 1.4 Pseudovector If @ is a pseudovector of a second order tensor $ , examine the nature of in consideration of Eq, (1.1.28). -eS a Ans, Decompose $ into the symmetric part and skew-symmetric part as Q 0 2-51 Hoan 0 2 en) [0 a -o 0 @ -a 0 3) Namely vector a is expressed with components @,, @; and @, as =a +80 +60 “ where « is obtained by —£-S (see Eq, (1.1.26) for example), this implies that components of the skew-symmetric part of the second order tensor are composed of components of the pseudovector. If $ is assumed to be a ve- locity gradient Vu, the vorticity vector « is derived from components of the spin tensor, which is the skew-symmetric part of the tensor Vu Exercise 1.5 Material Objectivity ‘The upper convective Maxwell model constitutive equation (a linear vi coelastie model, in Chapter 7) can be written by THAT=Ime @34 1 Fundamentals in Continuum Mechanics where + is second order tensor, 4 and 7p are constant and e is the rate of deformation tensor (see Eq. (1.1.16)). Show that Eq, (1) satisfy the ma- terial objectivity Ans. ‘The material objectivity has to be met by the linear transformation for tensor Aas follows A=Qlt) APQU) Q) where Q(t) is a rotation tensor, defined in Eq. (1.2.1). Therefore, the con- stitutive Equation of Eq, (1) has to be invariant by the transformation of Eq, (2), that is aeAt ape @) where P= QQ" @ F=Q4Q" ° and, e'=QeQr o knowing that scalar constants are frame invariants, In order to verify the invariance of Eq. (1), take the transformation to the model equation poe Q[+442=2me] Qr om \ Since Eq, (7) is linear, we can write (0707)+-[049" }-2n(000") ®) ‘Thus we can recover the given equation by knowing Eqs. (4), (5) and (6) vear'=2me’ 0Exercise 35 In deriving Eq, (9), we used the principal of frame invariance (the ma- terial objectivity) for the upper convective derivative given in Eq. (1.3.16). Exercise 1.6 Reynolds’ Transport Theorem G rate of cl ea physical picture of the Reynolds’ transport theorem, considering the inge of a certain quantity F of matter moving through a control fed in Fig, 1.10, Ans, Consider a control volume of region A, which contains a quantity of matter at some times, indicated by the solid line. At some time later time +A‘, the boundary of the system has a new physical location as shown by the dotted line, at which the control volume occupies regions B and A minus C. The increment of the matter is written Am={m,(¢+At)—m, (t+ A1)+m,(¢4 A2)}=m, (0) w Region A at ime() Region B at timel +1) Region € at timelt +) Stream lines Fig, 1.10 System of moving control volume and fixed control volume Taking differentiation to Eq. (1) with respect to time At and after rear- rangement, we can write ‘am _mileeAs)=mil) , malt Atmel At) At At At ®36 1 Fundamentals in Continuum Mechanics The rate of change of m is calculated by taking the limit of Eq. (2) as At—0. The first term on the right hand side of Eq. (2) thus becomes ton Mallat) —mule) agp At mina [rv @) Since region A is fixed in the coordinates, we can write SE favs [Ew w Ina similar manner as the second term becomes vim Mall + At)~me((+A0) Ao At maine o ring is the rate of change of m through surface area S, from control volume (region A ), and which is expressed by surface integral Jity = fm as o where Fu is the flux of # (F is transported through the surface of the control volume by stream of flow with velocity 1). Similarly me is expressed tite = ~ [rw ds 0 where the minus sign means the inward to the surface of the control vol- ume. With Egs. (6) and (7), Equation (5) becomes tity =e = [rwas @) ‘Therefore, the rate of change of m is altogether written Jim S82 Sear Dif w= [fu [reas Oy This is a Lagrangian-to-Eulerian description of the rate of extensive integral quantity given by Eq. (1.5.10). ange of anExercise 37 Exercise 1.7 Principal Axes of Stress The pressure p defined in Eq. (1.6.17) is the mean of the normal stresses, that is, one third of the trace of the total stress tensor TT. It appears that is meant to be the mean of the principal stresses. Give the definition of the principal stresses and its direction of the principal axis, Ans When the direction of a stress vector f) is equal to that of an unit normal vector ft, if ta) is derived from a stress tensor T by Cauchy formula, the dir ction of # is called the direction of the principal f= T =A a) where 4 is 1r quantity. Equation (1) gives a relationship written as a(t 2 and Eq, (2) has to satisfy |tAl @) for the condition of f #0. Equation (3) is called the characteristic equa- tion of stress tensor T. Roots of Eq, (3) give eigenvalues, which are the principal stresses for the principal axes. The perfect fluid given by Eq. (1.6.17) is an isotropic fluid in a sense that a simple direct stress acting in it does not produce a shearing deforma- tion. In the functional relation between stress and deformation must be in- dependent of the orientation of the coordinates system. ‘The component form of Eq, (3) is written by T-4 Ta Ts A Ty |=0 @) Ty -2 with which we have a third order polynomial equation for A as follows VR AN2 + LA-1s=0 )38 1 Fundamentals in Continuum Mechanics ffi Iso independent of the choice of the Since A is independent of choice of the coordinates system, the cients of /1,/; and Js in Eq. (5) are coordinates. Therefore, /,,1, and Js are frame invariants. Equation (5) is also true for other tensors such as the rate of deformation tensor, which is dis ised in more detail in Chapter 7, /), J and Jy are respectively given by © a bade (s) Note that if o,ozand 0 are the principal stresses, [,,1, and J, are re- spectively given by h=ou= Yon =ai+or+03 (9) (oe y yo) )=(oo2 +0203 +010) (10) and Is 2detoy = o10203 ay Problems 1-1, Show that Q-Q” is a skew-symmetric tensor, and whose pseudovee- toris © 2.07 40.07 0070 na0-0" =-0.07 =-(0.0" 7 Ans.| fij=Ob)ry (iid raion 2 Se fed) r= 00+-7=(0-.0" br=nxrProblems 39 1-2. If Q is a rotation tensor, show that Q is @ unitary matrix, which satis- fies Q” =Q~'. Consider the two dimensional axis rotation by 0 where the frame is transferred x'—y’ to. x—y “co et 1-3. Proof for Cartesian vectors w and v, that Vx(uxv)ov-Va—u. Vv ulV-v)—o(V-a) Ans. 1-4, When a scalar function p(x.) differs form a material surface, but p(x.t) moves with a velocity v different from the stream velocity w, show that 1F plese) the material surface, Pw .up, aoa Ans, J but move with relative velocity w= a=» std not with material pin Bo (u=v}-¥p=(0-») se = (unr vp= (uv)40 1 Fundamentals in Continuum Mechanics 1-5. Verify that the Finger tensor defined in Eq. (1.4.4), ie.C’ is symmetric. Consider E'" in the Cartesian coordinates i 1-6. Obtain components 7,,, T,, and T., of tensor T. 1-7. Show that the velocity gradient tensor Var does not satisfy the princi- ple of frame invariance. Ans. Vay QT 0.07 Van QT +02 vu" =Q-Va-Q" +s not near wansforesion of Q Nomenclature A: material parameter @ acceleration © = Cauchy strain tensor (C+: Finger tensor E__: displacement gradient tensor rate of strain tensor unit base vectorBibliography 41 BROT een L, woes eepee unit dyad force body force gravitation acceleration unit tensor Jacobian ‘material line element Intermediate scale large scale ‘molecular scale unit normal vector pressure rotation tensor ‘general second order tensor surface area total stress tensor stress vector time velocity position vector in vector space Knudsen number Kronecker delta (polyadic) alternator Eddington notation eigenvalue scalar potential angular velocity spin tensor vorticity vector Bibliography ‘The content in this chapter is standard matter and treated in almost all texts, ‘The mathematic methods in accounting Cartesian vectors and tensors are found in 1 A. Rutherford, Vectors, Tensors, and the Basie chanics, Prentice iquations of Fluid Me- Hall, Inc., Englewood Cliffs, NJ, 1962.42 1 Fundamentals in Continuum Mechanics 2. F.B. Hildebrand, Advanced Calculus for Applications (2nd Edition), Prentice-Hall, Inc., Englewood Cliffs, NJ, 1976. 3. R.B. Bird, C. F.Curtiss, R. C. Armstrong and O. Hassanger, Dynamics of Polymeric Liquids 2nd Edition), John Wiley & Sons, A Wile Interscience Publication, New York, Vol. | and Vol. 2, 1987. Appet dices are useful 4, G.B. Arfken and H. J. Weber, Mathematical Methods for Physicists (6th Edition), Elsevier, Academic Press, Amsterdam, 2005, Some inte ‘ng topics on microhydrodynamies are found in 5. 8. Kim and S. J. Karrila, Microhydrodynamics, Principles and Selected Applications, Butterworth-Heinemann, a division of Reed Publishing (USS.A,) Inc., Oxford, London, 1991 6. P. Tabeling, Introduction to Microfluidics, Oxford University Press Inc., New York, 2005,