Chapter 4 Dynamic Stability 4.

1 Introduction In static stability analysis we examined the moments brought about immediately after the disturbance. However a system is said to be dynamically stable if it finally returns to the equilibrium position. Hence to examine the dynamic stability we must analyze the subsequent motion. The motion following an intended control input or a disturbance is called response. Obtaining the response in the case of airplane is an involved task. In this introductory course we would obtain the conditions which ensure dynamic stability. The topic of response would be illustrated in chapter 5 with examples of simpler systems. 1

The treatment of dynamic stability is divided into the following topics. 1. Derivation of the equations of motion, in vector and scalar forms, for a rigid body. 2. Axes system used for describing airplane motion. 3. Equations of motion in stability axes system. 4. Equation of motion with small perturbation. 5. Linearization and decomposition of small perturbation equations. 6. Stability derivatives. 7. Solution of the equations of motion for longitudinal motion. 8. Equations of motion in state variable form
2

9) Stability diagrams 10) Approximations of longitudinal motion 11)Lateral dynamic stability 4.2 Equations of motion The equations of motion are obtained by applying the Newtons second law to the motion of airplane. For this purpose the airplane is treated as a rigid body which is translating as well as rotating. This motion is decomposed as:

(a) translation of the c.g. of the airplane with reference to an inertial frame which is taken as a frame fixed at a point on the earth and (b) rotation with respect to the inertial system of a body axes system, attached to the airplane. The linear velocity, vector and the angular velocity vector are resolved along the body axes system (see Eqs.4.12a and 4.12b). It may be pointed out that when a body axes system is used, the moments of inertia and products of inertia calculated with respect to this system would remain constant except for changes due to consumption of fuel by the engine or deflection of control surfaces.

For applying Newtons second law we need an expression for the acceleration of an elemental mass dm located at a point on the airplane. An expression for it (acceleration), in context of the aforesaid decomposition, is derived in the next subsection.

4.2.1 Acceleration of a particle on a rigid body Let the fixed frame of reference be EXeYe Ze with origin at E. Let the body fixed coordinated system attached to airplane be (Fig.4.1) OXYZ. Let the unit vectors along OX, OY and OZ axes be i,j and k. It may be mentioned that the quantities in bold are vectors. With reference to the co-ordinate systems, the position vector of the elemental mass dm at point P on the body can be expressed as:

E P = E O + O P = E O + x i + y j+ z k ;

(4.1)
6

Fig 4.1 Motion of a rigid body (Adapted from Ref.4.1, chapter 1)

The velocity of an elemental mass dm the point P is the time derivative of its position vector i.e.
d (EP) d (EO) d (OP) d (EO) d ( x i + y j + z k) V= = + = + dt dt dt dt dt dx dy dz di dj dk = V0 + i + j + k + x + y +z dt dt dt dt dt dt d (EO) V0 = dt

(4.2) (4.3)

For a rigid body, the distance between any two points is constant. Hence (4.4) dx/dt = dy/dt = dz/dt = 0. When a system of axes rotates with angular velocity , then d i dj dk (4.5) = i ; = j; = k

dt

dt

dt

dOP = ( xi + yj + zk ) = OP dt

Thus, we can write the velocity of the elemental mass at point P as: (4.6) V = V0 + ( x i + y j + z k)

= V0 + OP
d V d ( V0 + OP ) a= = dt dt d V0 d d OP a = a0 + OP + ; a0 = dt dt dt

(4.7) (4.8) (4.9)

The acceleration is the rate of change of velocity i.e.

d OP Replacing by x OP we get the acceleration (a) dt of the elemental mass at point P as:

a = a0 + OP + ( OP)

9 (4.10)

Applying Newtons second law of motion, we get the equations motion is vector form as:

F = a dm
m

(4.11)

M = OP a d m
m

(4.12)

4.2.2. scalar form of equations of motion Though the vector form of the equations of motion is compact, we need the equations in scalar form to solve them. This is done by writing the expressions for acceleration, forces and moments as sum of their x-, y-, and z- components and then equating the expressions for the three components on the 10

two sides of Eqs.(4.10) and (4.11). While deriving the scalar form some terms are not resolved into components, since they would vanish on integration over the mass. Let the origin on the body have a velocity (Vo) i.e. V0=ui + vj+ wk (4.12a) and angular velocity , =pi + qj+ rk. Then, the velocity of any point on the body as (4.12b) shown earlier is: V=V0+ x OP. The acceleration of the point P , from Eq.(4.9), is :

d (u i + v j + w k ) d d OP a= + OP + dt dt dt

(4.13)
11

Expanding the terms on the right hand side of Eq.(4.13) gives:

a = u i + v j + w k + u ( i) + v ( j) + w ( k) +{ p i + q j + r k + p ( i) + q ( j) + r ( k) } OP + ( OP ) (4.14)
i i i i i i

Equation (4.11) can now be written as:

F=
i m

{u i + v j + w k + u( i) + v( j) + w( k)}dm
i i m

+ p i + q j + r k + p( i) + q( j) + r ( k) OP dm + ( OP dm)
m

(4.15)

Taking O as the center of mass, we have,

m

OP dm = 0

(4.16) 12

Hence Eq.(4.15) simplifies to:

F = m [u i + v j + w k + u(r j q k) + v (pk r i) + w (qi pj)] = m (u+ qw r v ) i + m (v + r u pw) j + m (w + pv qu) k (4.17)
i i i i i i

We can express F as:

F = Fx i + Fy j + Fz k
Equating Eqs.(4.17) and (4.18) we get the force equations in scalar form as:
i

(4.18)

Fx = m(u + qw rv ) Fy = m(v + ru pw) Fz = m(w + pv qu)

i i

(4.19)
13

Now, we derive the scalar form of moment equations. The equation is:

M = OP a dm
m

M=

This equation involves several vector products. We obtain their components in the following steps. = pi + q j + r k Differentiating this expression with time and using Eq.(4.5) we get: i i i i = p i + q j + r k + p ( i) + q ( j) + r ( k)

OP [a0 + OP + ( OP)]dm

(4.20)

= p i + q j + r k + p (r j qk) + q( pk r i) + r (qi pj) = pi + q j + r k

i i i

(4.21)
14

 OP = ( pi + qj + rk ) ( xi + yj + zk )
= ( qz ry )i + ( rx pz ) j + ( py qx )k
Similarly using Eq.(4.21) we get: (4.22)

OP = ( q z r y ) i + ( r x p z ) j + ( p y q x ) k

(4.23)

( OP) = (pi + q j + r k) {(q z r y)i + (r x pz) j + (py q x)k} = (pqy q2x r2x + pr z)i + (qr z r2y p2y + qpx) j + (r px p2z q2z + r qy)k OP ( OP) = (py q xy r x z + p z2)i + (qz r yz pyx + q x ) j + (r x p zx q zy + r y2)k (4.25)
15

(4.24)

O [( O )] = {(pr xy p2zy q2zy + qr y2) P P (pqxz p2zy r2zy + qr z2)} + {(qpyz q2xz r2xz + r pz2) i (qr yx q2xz p2xz + r px2)}j + {(r qzx r2yx p2yx + pqx2) (r pzy r2yx q2yx + pqy2)} k
Let (4.26) (4.27)

M = M xi + M y j + M zk

Using Eq.(4.16), Eq.(4.20) simplifies to:

M = OP ( OP ) dm + OP ( OP ] dm
m m

(4.28)

16

OP ( OP)dm = [( p y
m i 2 i i i 2 i i i i

q xy r xz + p z )i +
2 i 2 i i i

( q z r yz p yx + q x ) j + ( r x p zx q zy + r y2 )k]dm = ( A p F q E r )i + ( B q Dr F p) j + (C r E p Dq)k (4.29)

Where
i i i i i

A = I xx = ( y 2 + z 2 ) dm, B = I yy = ( x 2 + z 2 ) dm;
m m

C = I zz = ( x 2 + y 2 ) dm , D = I yz = I zy = yz dm;
m m

(4.30)

E = I zx = I xz = xz dm , F = I yx = I xy = xy dm;
m m
17

OP[(OP)]dm = [{( prxy p2 zy q2 zy + qry2 )

m 2 2 2 2 2 2

( pqxz p zy r zy + qrz )}i +{(qpyz q xz r xz + rpz ) (qryx q2 xz p2 xz + rpx2 )}j +{(rqzx r2 yx p2 yx + pqx2 ) (rpzy r2 yx q2 yx + pqy2 )}k]dm
= [qr ( y 2 z 2 )dm + Fpr D(q 2 r 2 ) Epq ] i
m

+[rp ( z 2 x 2 )dm + Dqp E (r 2 p 2 ) Fqr ] j

m

+[ pq ( x 2 y 2 )dm + Erq F ( p 2 q 2 ) Drp ] k

m

(4.31)
18

We simplifying Eq.(4.31) by noting that:

(y (z (x

z )dm = [( y + x ) ( z + x )]dm = (C B)
2 2 2 2 2 m

x )dm = [( z + y ) ( x + y )]dm = ( A C)
2 2 2 2 2 m

y )dm = [( x + z ) ( y + z )]dm = (B A)
2 2 2 2 2 m
2

OP [ ( OP)]dm = [(C B)qr + Fpr D(q

r )
2

Epq]i + [( A C ) rp + Dqp E (r 2 p 2 ) Fqr ] j +[( B A) pq + Erq F ( p 2 q 2 ) Drp ] k

(4.32)
19

Hence M = ( A p F q E r )i + ( B q D r F p) j + (C r E p D q )k + [(C B )qr + Fpr D(q 2 r 2 ) Epq ] i + [( A C )rp + Dqp E (r p ) Fqr ]j

2 2 i i i

+[( B A) pq + Erq F ( p 2 q 2 ) Drp ]k

Using Eq.(4.27) in Eq.(4.33) yields:

(4.33)

M x = A p ( B C ) qr + D ( r q ) E ( pq + r ) + F ( pr q )
2 2

(4.34) (4.35)

M y = B q (C A)rp + E ( p r ) F (qr + p ) + D(qp r )

2 2 i 2 2 i i

M z = C r ( A B ) pq + F (q p ) D(rp + q ) + E (rq p ) (4.36) 20

4.2.3 Forces acting on the airplane The external forces acting on an airplane are the thrust (T), the aerodynamic forces (A) (lift, drag and side force) and the gravitational force (mg). In vector form, we can write Eq.(4.11) as,

T + A + mg = a dm

(4.37)

It may be recalled that we have made the following Assumptions. (a) The airplane is rigid. (b) The reference frame attached to the earth is a Newtonian frame. (c) Flat earth model is used for gravitational force.
21

Before obtaining the scalar form of Eq.(4.37) the following should be noted. (a) The thrust vector acts roughly along the fuselage reference line (FRL). (b) The aerodynamic forces are resolved such that the drag is parallel to the free stream direction and the lift and the side force are in mutually perpendicular directions to the free stream. (c) The gravitational force acts vertically downwards . (d) To obtain the scalar form of Eq.(4.37), T,A and mg must be expressed in one coordinate system. Towards this we define different coordinate systems and obtain relationships between them .
22

4.3 Axes systems We now examine various types of coordinate systems used in flight dynamic analysis. We will assume that the aircraft has a plane of symmetry. The various types of axes systems are: 1. Ground axes system (EXeYeZe) 2. Local horizon system (OXhYhZh) 3) Body axes system (OXbYbZb) and special cases namely wind axes system (OXwYwZw) and stability axes system (OXsYSZs) Some of these axes systems are shown in Fig.4.2 and are defined as follows.
23

Fig 4.2 Axes systems (Adapted from Ref.4.1, chapter 4)

24

1. Ground axes system: The system is fixed with respect to earth and defined as follows. The origin E is a point on the earths surface. The EZe-axis is vertical and positive downwards. The EXeand EYe-axes are contained in a horizontal plane and are directed in such a way that EXeYeZe forms a right handed system (Fig.4.2). Remark: Reference 1.8, chapter 4 calls the aforesaid ground axes system as navigational system. The point E is chosen such that it is beneath the c.g. of the airplane at the start of the flight. The axis OXe could be towards the local north and the axis OYe towards 25 the local east. It may be added that there are other

definitions of navigational system. 2.Local horizon system: This system has its origin at a point O on the airplane which lies in the plane of symmetry. The axes OXh, OYh and OZh are parallel to EXe, Eye and EZe respectively (Fig.4.2). 3.Body axes system: Any axes system fixed to the airplane and moving with it is called body axes system. An axes system used for prescribing the moments and products of inertia of the airplane and denoted by OXbYbZb is as follows. The axis OXb is contained in the plane of symmetry, and is positive forward. Generally it coincides with the fuselage reference line (FRL). OZb is perpendicular to OXb and 26 is contained in the plane of symmetry. OZb is

positive downwards for normal flight attitude of the airplane. OYb is perpendicular to the plane of symmetry and is directed in such a way that OXbYbZb is a right handed triad (Fig.4.2). Wind axes system: This is also a body axes system in which the axis OXw is tangent to the flight path in the undisturbed state and positive in forward direction. OZw is perpendicular to OXw and contained in the plane of symmetry and positive downwards for normal flight attitude of the airplane. OYw is perpendicular to both OXw and OZw and is directed in such a way that OXwYwZw is a right handed system (Fig.4.2).
27

Stability axes system : This is also a body axes system and denoted by OXsYsZs . The axis OXs lies in the plane of symmetry and if the undisturbed flight is with no sideslip (=0), then it (OXs axis ) points in the flight direction. If 0 then OXs coincide with the projection, in the plane of symmetry , of the flight velocity vector . The axis OYs is perpendicular to the plane of symmetry, positive in starboard direction , and the axis OZs points downwards such that OXsYsZs is a right handed system. It may be pointed out that the angle between OXb and OXs is the angle of attack. If =0 , then drag and lift are in directions opposite to OXs and OZs axes respectively. 28

4.3.1. Relationships between the various axes systems Relation between ground axes and local horizon systems: Since the ground axes system and local horizon systems are parallel to each other, the unit vectors in the two systems are related as: ih = ie, jh = je, kh = ke. In matrix form the relationship can be expressed as: ih ie

jh k h

= je k e

(4.38)

Relation between body axes and local horizon systems: In general, the transformation between these two axes systems can be written as:

29

i b = a11i h + a12 jh + a13k h jb = a21i h + a22 jh + a23k h k b = a31 i h + a32 jh + a33k h

The set of Eqs.(4.39) indicates that nine quantities namely, a11,a12,.,a33 are needed to prescribe the transformation. However we have the following six relationships between the direction cosines: ib.ib = 1, ib.jb = 0, ib.kb = 0; jb.jb = 1, jb.kb = 0; kb.kb = 1. Thus, only three quantities are needed to describe the transformation from one coordinate system to another. There are various ways to arrive at the three
30

(4.39)

quantities. In the approach called Eulerian angles, we rotate one coordinate system through three angles about three different axes and arrive at the new system. This approach is followed in flight mechanics. The steps are as follows. I) First, rotate OXhYhZh system about OZh through yaw angle to get a new coordinate system OX1Y1Z1 (Fig.4.3). The unit vectors in this system are i1 , j1 and k1. The two sets of unit vectors are related as: i1 = cos ih+ sin jh j1 = -sin ih+ cos jh k1 = kh.
31

Fig 4.3 Successive rotations of local horizon system 32 (Adapted from Ref.4.1 chapter 4)

In matrix notation, these relations can be written as:

II) Next, rotate OX1Y1Z1 system through pitch angle about OY1 axis to get OX2Y2Z2 (Fig.4.3). These two axes systems are related as:

i1 cos sin 0 ih j j1 = sin cos 0 h k 0 0 1 kh 1

(4.40)

i 2 cos j2 = 0 k sin 2

0 sin i1 j 1 0 1 0 cos k1

(4.41)

III) Finally, rotate OX2Y2Z2 system through bank angle about OX2 to get OXbYbZb. These two axes33 systems are related as:

0 i b 1 jb = 0 cos k 0 sin b

0 i2 sin j2 cos k 2

(4.42)

Combining Eqs.(4.40),(4.41) and (4.42), we get:

0 cos 0 sin cos sin 0 ih ib 1 0 0 1 0 sin cos 0 j jb = 0 cos sin h (4.43) k 0 sin cossin 0 cos 0 0 1kh b
After multiplying the matrices in Eq.(4.43), the relationship between ib,jb and kb and ih,jh and kh is presented in Table 4.1, which also shows relationship between ih,jh and kh and ib,jb and kb. For example: kh=-sin ib+ sin cos jb+ cos cos kb . Remark: The angles , and are the Eulerian angles.
34

ib ih

Jb

kb cos sin cos + sin sin

cos cos sin sin cos - cos sin

jh cos sin

sin sin sin +cos cos

cos sin sin -sin cos

kh -sin

sin cos

cos cos

Table 4.1 Relation between direction cosines of wind 35 axes and local horizon systems

Relation between OXwYwZw and OXbYbZb systems: we describe this relationship for the sake of completeness. We must note that the OXb and OZb axes are contained in the plane of symmetry of the airplane. Hence, only two angular rotations namely the sideslip angle and the angle of attack are needed to determine the relative orientation of the two sets of axes. I) First, rotate OXwYwZw through an angle - about the OZw axis to obtain the OX3Y3Z3 system (See Fig.3.1for directions of OXb and V) Thus, we have
i 3 cos j3 = sin 0 k 3 i sin 0 w cos 0 j w 0 1 k w

(4.44)
36

II) Then, rotate the OX3 Y3 Z3 system through an angle about the OY3 axis to get the body axes system. Thus, we have

ib cos 0 sin i3 j 0 3 jb = 0 1 k sin 0 cos k b 3

Combining Eq.(4.44) and (4.45), we have:

(4.45)

ib cos 0 sin cos sin 0 iw sin cos 0 j jb = 0 1 0 w k sin 0 cos 0 0 1kw b

The relation between direction cosines of wind axes and body axes systems are presented in Table 4.2 .

(4.46)

37

ib iw cos cos

jb sin

kb sin cos

jw kw

-cos sin -sin

cos 0

-sin sin cos

Table 4.2 Relation between direction cosines of wind axes and body axes systems Remark: It may be noted that the angles and are the angle of attack and the angle of sideslip. 38

Thus, we have obtained the relationships between the various axes systems used in flight dynamics. We now obtain the equations of motion in stability axes system. 4.4 Equation of motion in stability axes system As mentioned earlier the scalar form of the equations of motion is obtained by resolving all the forces and moments along one axes system. Stability axes system is used for this purpose. It is assumed that in the equilibrium state the airplane is in an unaccelerated climb at angle 0 to the horizontal. When stability axes system is used, the axis OXb is in the plane of symmetry but in the direction of flight. The following may be noted . 39 (a) When the airplane encounters a disturbance

it (airplane) is displaced but the orientation of the stability axes remains fixed in space. (b) If u,v,w are the components of the velocity vector along the stability axis then the components in the undisturbed flight , u0,v0 and w0 are respectively u0= |V| v0 =0 and w0=0. (c) The moments of inertia of the airplane may be calculated about some convenient axes system . These will have to be converted when stability axes are used (see for example Ref.4.3, appendix C). The transformation rules are given in the previous section. Let X,Y and Z be the components of the aerodynamic and propulsive 40

forces along the OXs, OYs and OZs axes respectively. The gravitational force is mgkh. Noting that kh = -sin is + cos sin js + cos cos ks, The components of the gravitational force along OXs, OYs and OZs axes are (Fx)gravity = -mg sin (Fy)gravity = mg cos sin (Fz)gravity = mg cos cos Let L, M and N be the moments due to the aerodynamic and propulsive forces about the OXs, OYs and OZs axes. Using Eqs.(4.19), (4.34) to (4.36) and (4.47) we get the following equations of motion in the scalar form.
41

(4.47)

They are actually the components of forces and moments along OXw, OYw and OZw axes. It may be noted that for an airplane with plane of symmetry two of the products of inertia namely D and F are zero.

m (u + qw rv ) = X mg sin m (v + ru pw) = Y + mg cos sin m ( w + pv qu ) = Z + mg cos cos

i i

(4.48) (4.49) (4.50) (4. 51) (4.52)

42 (4.53)

I xx p I xz r + ( I zz I yy )qr I xz pq = L ' I xx q + ( I xx I zz )rp + I xz ( p 2 r 2 ) = M I zz r I xz p + ( I yy I xx ) pq I xz qr = N

i i i

Remarks: i) As noted earlier u,v and w are the components of velocity along the stability axes systems. In undisturbed flight these components are denoted by u0,v0 and w0. Since stability axes system is used, v0=w0=0. ii) Equations (4.48) to (4.53) are the six equations for the six unknown quantities viz. u,v,w, p, q and r. It may be recalled that a rigid airplane is a system with six degrees of freedom. These six equations form the set of governing equations for the dynamics of rigid airplane. 4.5 The small perturbation theory 4.5.1 Introduction to small perturbation theory 43 The small perturbation theory is used commonly to

analyse the stability of various phenomena. The basic ideas are as follows . I) Each system is described by a certain number of variables called the state variables and by a set of equations. For example in the case of the motion of rigid airplane, the state variables are u,v,w,p,q,and r and it is governed by six equations namely Eqs.(4.48) to (4.53). The state variables will have certain values in the undisturbed state. After disturbance these values will change and the deviations will be functions of time. II) We assume that the deviations in the state variables, caused due to a disturbance, are small compared to their values in the undisturbed state. For example if a 44 variable f had a value f0 in undisturbed state, it

becomes f0+f after the disturbance. Then f << f0 and is a function of time. The quantity f is called the perturbation in f about f0. III) We substitute in the governing equations, (f0+ f) in place of f and likewise for the other state variable. Then we get equations for fs. As the disturbance is small the products of the deviations or their second and higher powers are ignored and equations for fs become linear. This will be clearer when we take up the motion of the airplane. IV) We solve these equations for fs and examine the evolution of fs with time. If all fs go to zero ultimately, then the system is stable one. If anyone of 45 the fs does not ultimately go to zero then the

system is unstable in that variable. See Fig.1.6 for six possible ways in which f may change with time.

46

4.5.2 Application of small perturbation theory to motion of airplane As regards the problem of the stability of an airplane, the application of small perturbation theory can be carried in the following steps. I) we assume that u, v, w, p, q, r, X,Y,Z, L, M, N and the control deflections a , e and r have the values u0,v0,w0,p0,q0,r0,X0,Y0,Z0, L0, Mo,No , a0 , e0 and r0 respectively in the equilibrium state; some of these values could be zero. After the disturbance we express these quantities as:
47

u = u0 + u; p = p0 + p;

v = v0 + v; q = q0 + q;

w = w0 + w; r = r0 + r; (4.54)

X = X0 + X; Y = Y0 + Y; Z = Z0 + Z; L = L0 + L; M = M0 + M; N = N0 + N; a = a0 + a; e = e0 + e; r = r0 + r; =0+; 0 is the angle of climb.

Note:1) An airplane in steady climb is also in an equilibrium flight. The angle of climb, here denoted by , was denoted by in performance analysis. 2 (a) The quantities u, v, and w are much smaller as compared to u0 i.e. u, v, w <<u0. (b) The quantities (l p), (l q), (l r)<< u0, where l is a reference length; p, q, r have
48

been multiplied by l so that the product has dimensions of velocity. (c) X, Y, Z are small as compared to the lift developed by the airplane. 3) When reference flight is steady and in the plane of symmetry, then v0 = p0 = q0 = r0 = 0 = 0 =D=F= 0 . where 0= bank angle and 0= angle of yaw. 4) Further, if the X-axis is the same as Xs-axis, then the velocity component w0 is zero. II) To obtain the equations for u, v, w, p, q and r, let us introduce the perturbation quantities given by Eq.(4.54) in the equations of motion i.e. 49 Eqs.(4.48) to (4.53).

Let us consider Eq.(4.48). With introduction of perturbation quantities as given in Eq.(4.54), it becomes:

d(u0 + u) m[ + (q0 + q)(w0 + w) (r0 + r )(v0 + v)] dt = X0 + X mg sin(0 + ) (4.55)

Note: w0 = v0 = q0 = r0 = 0. We would get five more equations by substituting perturbation quantities from Eq.(4.54) in Eqs.(4.49) to (4.53) . These five plus Eq.(4.55) are the six equations for u, v, w, p, q and r. These are non-linear equations as they involve products of 50 unknowns viz. q x w, r x v etc.

4.5.3 Linearization of equations for perturbation quantities As mentioned in the previous paragraph the equations for perturbation quantities are non-linear. They can be linearised by ignoring, the products involving unknown quantities, as small. However the linearization process can be better appreciated by writing the equations in non-dimensional form. We use the following nondimensionalization scheme.

u v w lq lr tu0 (4.56) ;q = ;r = ; t = Let u = ; v = ;w = u0 u0 u0 u0 u0 l

51

Remark: It may be pointed out that for the purpose of nondimensionalization, we use (a) u0, the flight velocity in undisturbed flight, as the reference velocity, (b) l, as the reference length; it can be lt or b as the case may be. (c) as regards the reference time following can be mentioned. The reference quantities l and u0 arise as different airplanes would have different lengths and would fly at different velocities. However, there is no imposed time scale in the stability problem. Hence the particular variations of u etc. with time, obtained for an airplane, are the result of the length and velocity scales for the airplane. Hence the reference time (tref) is to be 52 obtained from a combination of l and u0 i.e.

^ u , q by u 0 and so on in Replacing u by u 0 q l Eq.(4.55) , it takes the following form:

u u d (u 0 + u u 0 ) 0 0 + ( q )( w u0 ) ( r )( v u0 )] m[ l l l d (t ) u0

tref=l/u0. Consequently the non-dimensional time ( ^ ) in Eq.(4.56) is given as: t ^ t/t =tu / l. t=
ref o

= X 0 + X mg sin ( 0 + )
2

(4.57)

mu0 d (1+ u) Or + q w r v] = X 0 + X mg sin(0 + ) (4.58) [ l dt 53

Dividing Eq.(4.58) by mu02/l we get the nondimensional form of Eq. (4.55) as:
d (1 + u )

d t

l X 0 l X g l + q w r v = + 2 sin ( 0 + ) (4.59) 2 2 mu0 mu0 u0

^ ^ ^ ^ In Eq.(4.59) the products q w and r u appear. Now we note that though u, v, w, q, r etc. ^ may take some arbitrary values, the quantities u, ^ ^ q, r etc. are, by definition, much v, w, ^ ^ smaller than one or infinitesimal of first order. Hence ^ ^ u, q etc. are not negligible but the products like ^ ^ ^ ^ qw , r v are much smaller i.e. second order
54

infinitesimal or roughly speaking of the order of 0.01. Hence these products are negligible as compared to 1. It may also be remembered that we are solving for u, ^ ^ v etc. and they cannot be ignored when they appear individually. Thus Eq.(4.59) reduces to:

d u

l X 0 l X g l = + 2 sin ( 0 + ) 2 2 mu0 mu0 u0 dt

(4.60)

Further,

sin( 0 + ) = sin 0 cos + cos 0 sin

Though 0 may not be small, is small and hence cos 1 and sin = . Consequently, sin( 0 + ) sin 0 + cos 0 (4.61)
55

Reverting back to the dimensional form, Eq.(4.60) becomes:

We have considered the undisturbed flight as steady climb at 0 . It may be recalled from the performance analysis of climb flight that the equation of motion along OXs axis is: T0-D0-W sin 0 =0 where T0 and D0 are thrust and drag in steady climb. Noting that: X0 = Aerodynamic + propulsive force = T0-D0 and W= mg, we get: X0- mg sin 0 =0. 56 Hence Eq.(4.62) reduces to:

d u m = X 0 + X mg (sin 0 + cos 0 ) dt

(4.62)

d u = X mg cos 0 m dt

Or

m u = X mg cos0

(4.63)

To derive the small perturbation forms of Eqs.(4.50), (4.52),(4.49)(4.51) and (4.53), we can follow the steps adopted in deriving Eq.(4.63) from Eq.(4.48). The derivations are left as an exercise for the reader. Recalling from section 1.2.4, we simplify the analysis by splitting the six degree of freedom system into two systems each with three degrees of freedom namely longitudinal motion and lateral motion . The final form of the equations for longitudinal and lateral motions are given below and presented in Table 4.3. Some clarifications are given 57 as remarks after the table.

Small perturbation equations for longitudinal motion: i (4.63) m u = X mg cos 0

m (w u0 q) = Z mg sin 0 ,

(4.64) (4.65)

Iyy q = M

Small perturbation equations for lateral motion:

m(v+ r u0) = Y + mg cos0 Ixx p Ixz r = L' Ixz p + Izz r = N

i i i i

(4.66) (4.67) (4.68)

58

Equations of motion

Small perturbation form

Longitudinal motion
m(u + qw rv) = X mg sin
m(w+ pv qu) = Z + mg cos cos
i

m u = X mg cos 0
m(w u0q) = Z mg sin0
Iyy q = M
i

I yy q+ (Ixx Izz )rp + Ixz ( p r ) = M

2 2

Lateral motion
m(v+ ru pw) = Y + mg cos sin
Ixx p Ixz r+ (Izz I yy )qr Ixz pq = L '
I zz r I xz p+ ( I yy I xx ) pq I xz qr = N
i i

m(v + ru0 ) = Y + mg cos 0

I xx p I xz r = L '
I xz p+ I zz r = N
59
i i

Table4.3 Equations of motion and small perturbation form

Remarks: Following aspects are considered while deriving the small perturbation form of Eqs.(4.50), (4.52), (4.49), (4.51) and (4.53) . (i) Equation (4.50): (a) The term pv would become p v and would drop out as product of two small quantities. (b) The term qu would become q(u0+ u). This has two terms q x u0 and q x u. The term q x u drops out as product of small quantities. However q x u0 remains. (c ) In the undisturbed flight of steady climb at angle 0, the equation of motion is: L0-W cos 0=0, L0 = lift in undisturbed flight
60

which in the present notation becomes Z0+mg cos 0=0. Taking these aspects into account and noting that is zero in steady climb, the small perturbation linearized form is given as Eq.(4.64) and compared with equation of motion in Table 4.3. ii) Equation (4.52): (a) The term rp would become r p and can be neglected as product of two small terms. (b) The term (r2-p2) would become (r)2- (p)2 and can be neglected as powers of small terms. (c) In steady flight pitching moment is zero and hence M0 is zero . 61

The linearized small perturbation form is given as Eq. (4.65) and compared with Eq.(4.52) in Table 4.3. Instead of q we write q as the variable. iii) Equation (4.49): (a) The term ru would become r(u0+ u) . The term ru is neglected and the term ru0 is written as ru0. (b) The term pw becomes pw and is neglected. (c) In steady flight there is no sideslip and hence Y0 is zero. The linearized small perturbation form is given as Eq. (4.66) and compare with Eq.(4.49) in Table 4.3. Instead of r we write r as the variable.
62

iv) Equation (4.51): (a) The term qr becomes qr and is neglected. (b) The term pq become pq and is neglected. (c) The rolling moment (L) in steady flight is zero. The linearized small perturbation form is given as Eq. (4.67) and compared with Eq.(4.51) in Table 4.3. v) Equation (4.53) : The terms involving pq and qr dropout as negligible and N0 is zero in steady flight. The linearized small perturbation form is given as Eq. (4.68) and compared with Eq.(4.53) in Table 4.3.
63

4.5.4 Expressions for changes in aerodynamic and propulsive forces and moments The aerodynamic forces and moments and the propulsive force vary with u, v, w, p, q, r, a , e and r and their derivatives. According to Ref.1.1, chapter 3, Bryan, who gave the basic frame work of stability analysis in 1911, assumed that these forces and moments can be expressed as functions of perturbation variables. This can be . . expressed in the form of a Taylor series as: x (u, v, w, p, q, r, u, v, ..a , e, r, T); T is a parameter indicating engine setting.
x x x x x x x = u + v + ... + u + ...+ a + e + r + T u v u a e r T 64 (4.69) + higher order terms

However to preserve the linear nature of analysis we ignore the higher order terms. Further to avoid unnecessary complications we express X, Y, Z, L, M and N in terms of only a few quantities which directly affect them. Table 4.4 presents the quantities and the perturbation variables on which they depend i.e.
X X X X X = u + w + e + T u w e T

(4.70)

Z Z Z i Z Z Z Z = u + w + i w+ q + e + T ; q = d /dt (4.71) u w q e T w
M M M i M M M M = u + w + i w+ q + e + T u w q e T w

(4.72)
65

Y Y Y Y Y = v + p + r + r v p r r L ' L ' L ' L ' X L ' = v + p + r + r + a v p r r a N N N N N N = v + p + r + r + a v p r r a

(4.73) (4.74) (4.75)

Quantities Factors u, w, e, T X . u, w, w , q, e, T Z . u, w, w , q, e, T M v, p, r, r Y u, p, r, r, a L u, p, r , r, a N Table 4.4 Changes in aerodynamic forces and moments and their dependence
66

Remarks: i) The simplification of expressing X,Z, ,N, in terms of only a limited number of variables is possible because we make the following reasonable assumptions. (a) X,W and M are affected only by the variables . of longitudinal motion i.e. u, w, w , q, e, T . . The dependence of X on w is ignored (Eq.4.70). (b) Y,L and M are affected only by the variables of motions out of the plane of symmetry viz. v, p, r and control deflections e, a. ii) These assumptions are valid for conventional airplanes with plane of symmetry having high aspect 67 ratio wings (A>4) and operating at moderate

angles of attack. See Ref.1.8 for treatment of airplanes with low aspect ratio wings and operating at high angles of attack . iii) The quantities X / u, X / w N/ r, N/ a are called stability derivatives. 4.5.5 Final form of small perturbation equations Substituting for X from Eq.(4.70) in Eq.(4.63) yields:
d u X X X X m = u + w + e + T mg cos 0 (4.76) dt u w e T

d X X X X w + mg cos0 = e + T Or (m )u dt u w e T

(4.77)
68

Let us introduce the notation:

Xu = 1 X 1 X 1 X 1 X , Xw = , X e = , X r = ; m u m w m e m T

(4.78)

Using this, we get

d ( X u )u Xw w + g cos 0 = X e e + X T T dt

(4.79)

In a similar manner let us introduce expression for Z, M, Y, L and N in Eqs.(4.71) to (4.75) respectively and use the following notations.
1 Z 1 Z 1 Z 1 Z 1 Z Zu = , Zw = , Zw = , Z e = , Z r = , (4.80) m u m w m w m e m T
Mu = 1 M 1 M 1 M 1 M , Mw = ,..., M e = , M t = (4.81) Iyy u Iyy w Iyy e Iyy t 69

1 Y 1 Y 1 Y Yv = , Yr = , Y r = m v m r m r

(4.82)

1 L' 1 L' 1 L' 1 L' 1 L' L'v = , L'p = , L'r = , L' a = , L' r = (4.83) Ixx v Ixx p Ixx r Ixx a Ixx r 1 N 1 N 1 N 1 N 1 N Nv = , Np = , Nr = , N a = , N = Izz v Izz p Izz r Izz a Izz r

(4.84)

Finally the small perturbation equations for longitudinal and lateral motion are as follows.

70

Longitudinal motion:

d ( X u )u Xw w + g cos 0 = X e e + XT T dt

(4.85)

d d Zu u + [(1 Z i ) Zw ]w [(u0 + Zq ) g sin 0 ] w dt dt = Z e e + Z T T (4.86) d d2 d Mu u (M i + Mw )w + ( 2 Mq ) w dt dt dt = M e e + MT T

(4.87)

71

Lateral Motion:
( d Yv )v (u0 Yr )r g cos 0 = Y r r (4.88) dt Ixz d d L 'v v + ( L 'p )p [ + L 'r ]r = L ' a a + L ' r r (4.89) dt Ixx dt Ixz d d + Np )p [ Nr ]r = N a a + N r r Izz dt dt

Nv v (

(4.90)

Remarks: i) The Eqs.(4.85) to (4.90) are the linearized small perturbation equations for longitudinal and lateral motions. Their solution would yield answers to two types of problems namely (a) response to a disturbance and (b) response to a control input.

72

In the case of response to a disturbance, with the control fixed, we assume that e , r , a , and r are zero. To give a disturbance we assume that one of the parameters like u, w,, r is given a small value at time t =0. Then we solve the equations to examine the changes, with time, in the value of the chosen parameter. This would give us information about dynamic stability. However it will be shown in section 4.7 that to deduce whether the airplane is stable or not, we do not need to solve the above differential equations. There is a simpler approach. In the case of response to control input, we assume that the control deflection is given as a function of 73 time and obtain the solution of the above equations.

For example we may prescribe that the elevator deflection e changes from zero to a value e1, in a small interval of time and then remains constant. The solution of the equations would give us the information about change in angle of attack and time taken to achieve the final value . Figure 4.4 shows a response wherein the airplane attains the final angle of attack after over-shooting the final value. In some cases the final value may be achieved without oscillation about the final value. However calculation of response is an involved task and some indication would be given in chapter 5. ii) Before solving the equations we need to know the stability derivatives. They are estimated in the next section.

74

(a) Elevator deflection

(b) Possible response of change in angle of attack.

Fig.4.4 Response to elevator deflection

75

4.6 Estimation of stability derivatives 4.6.1 Derivatives due to change of u These derivatives include X/u, Z/u and M /u. I) X/u : The changes in X are caused by changes in the drag and the thrust i.e.

X = D + T

(4.91)

Continuing with the linearized treatment of problem, the variation of X with u is expressed as:

D T X = u + u u u

(4.92)
76

X D T 1 T 2 = + = ( u 0 SC D ) + Or u u u u 2 u 1 T 2 C D = S (u 0 + 2u 0 C D ) + 2 u u

(4.93)

Recall that Xu= (1/m)(X/u) and let CDu= CD/(u/u0) As regards the term, T/u we note the following. (a) For gliding flight, T=0 and hence, T/u = 0. b) For a jet airplane, T is nearly constant over small intervals of u and hence, T/u = 0. c) For a piston engined airplane with variable pitch propeller the THP is nearly constant over a small range of u, hence: T = THP/u and consequently T/u = -THP/u2 = -D/u
77

As regards CDu we note the following: a) For subsonic flights with Mach number less than the critical Mach number, the drag coefficient remains constant with Mach number and hence CDu=0. b) When CD is a function of Mach number(M1), we can write: CDu=CD/(u/u0)=u0CD/(a0M1)=(u0/a0)CD/M1 =M1CD/M1; where a0= speed of sound under conditions of undisturbed flight. We use the symbol M1 for Mach number to avoid confusion with pitching moment (M). With the above considerations we can write:
Xu

u0 S 1 X 1 T = = (C D u + 2 C D ) + m u m u 2m

(4.94) 78

Xu =

2 u0 S

2mu0

(CDu

D ) + 2CD ) + 0 or ( mu0

(4.95)

CDQS QS 1 2 (CDu + 2CD ) + 0 or ( ); Q = u0 (4.96) = mu0 mu0 2

Or

QS (4.97) {(CDu + 2CD ) + 0 or (CD )} = mu0 Following Ref.1.1 , chapter 3 we introduce new quantities CX and CXu as follows: C X X 1 X CX = = ; C Xu = (4.98) QS (u / u0 ) (1 / 2) Su0 u

Consequently, we have:
2 Su0 1 X = Xu = C Xu 2mu0 m u

(4.99)
79 (4.100)

C Xu = {(CDu + 2C D ) + 0 or (CD )}

II) Z/u : The force in Z-direction is due to weight and the lift.

1 2 (4.101) Z = W u0 S CL 2 Ignoring the changes in weight during the disturbance we get:

CL Z 1 = S u 0 (C Lu + 2 C L ); C Lu = u 2 (u / u 0 )

(4.102)

Let C Z u Or Z u

1 Z = ; Th en C zu = (C Lu + 2 C L ) (1 / 2) S u 0 u
2 S u0 1 Z = = (C Lu + 2 C L ) m u 2 m u0

(4.103) (4.104)
80 (4.105)

C Lu

QS = (C Lu + 2 C L ) m u0 u0 C L C L C L = = = M1 (u / u0 ) a0 (u / a0 ) M1

It may be added that: a) At low subsonic Mach number CLu can be neglected. b) At sub-critical Mach numbers Ref 1.1, chapter 3 mentions that dCL / dM1 can be calculated using Prandtl -Glauert rule applicable to airfoils. However Ref.1.8, chapter 4 suggests that :
CLu dCL dCL = M1 = M1 dM1 dM1

(4.106)

The term CL as a function of Mach number is given as: 2 A C L = in rad 1 (4.107) 2 2 2

2+ A (1 + tan c / 2 ) +4 2 2 k

where A = Aspect ratio of wing, c / 2 = sweep of the mid-chord line, =(1-M12)1/2 and k=lift curve slope of airfoil/2.

81

Remark: The stability derivative CLu for Boeing-747 has been calculated at M1=0.8 in Appendix C. For this purpose CL is evaluated at M1=0.82 and M1=0.78. Using these two values dCL / dM is calculated at M1=0.8. III) M/u : Noting that
M = 1 2 u0 S c Cmcg 2

gives:

(4.108)

M 1 2 = ( S u0 c Cmcg ) u u 2
Cmcg 1 2 = S c (u0 + 2u0Cmcg ) 2 u But Cmcg=0 in equilibrium flight. Consequently,

82

CM M 1 = Scu0Cmu where Cmu = u 2 (u / u0 )

(4.109)

As shown earlier in Eq.(4.105) we get:

Cmu

Cm = M1 Where M1 is Mach number M1

Mu = 1 M QSc = Cmu Iyy u u0 Iyy

(4.110)

We define

(4.111)

In a manner similar to Eq.(4.106), we can write:

Cmu

Changes in Cm with M1 can be found out by obtaining Cm at nearby Mach numbers (see Appendix C) . The value is also affected by elastic bending of the wing and 83 fuselage.

Cm = M1 M1

4.6.2. Derivatives due to change of w These include X/w, Z/w and M/w. The discussion in this subsection is based on Ref.4.2, chapter 4. It may be noted out that in the undisturbed flight, Xs-axis is along the flight direction. i.e. V = u0i, w = 0, v = 0. After disturbance, the airplane has w (Fig.4.5). Thus the relative wind makes an angle = w/u0. The lift (L) and drag (D) are now perpendicular and parallel respectively to the relative velocity (VR) as shown in Fig.4.5. I) X/w : From Fig.4.5 , X = L sin D cos (4.112) X 1 X 84 = (4.113) w u0

Fig.4.5 Stability derivatives due changes of w (Sketch adapted from Ref.1.1, chapter 3)
85

X 1 L D = {L cos + sin + D sin cos }(4.114) Hence w u0

Since is small, cos = 1 and terms involving sin are ignored. 1 X 1 D (4.115) Hence Xw = = (L )

m w

mu0

1 1 1 2 2 = { u0 S CL ( u0 S CD )} mu0 2 2 QS = (CL CD ) = (CL CD ) (4.116) 2mu0 mu0 CX = CL CD where Cx Xw mu0 = QS

(4.117)
2 u0 S

Note: For Mach number less than the critical Mach number the drag polar is given by:

86

CD = CD0

2 CL 2C L . Hence C D = C L + Ae Ae

(4.114)

II) Z/w: From Fig.4.5:

Z = (L cos + D sin )

(4.115) (4.116)

Further Hence,

Z 1 Z = w u0

Z 1 L D = {L sin + cos + D cos + sin }(4.117) w u0

Taking cos =1 and ignoring the terms involving sin gives:

87

Z 1 L = { + D} w u0 L 1 Z 1 Zw = = + D) ( m w mu0 1 1 1 2 2 = { u0 S CL + u0 SCD } mu0 2 2 =

2 u0 S

(4.118) (4.119)

2mu0

(CL

QS (C L + C D ) + CD ) = mu0

(4.120) (4.121)

C Z = CL

Zw mu0 + CD = QS

88

III) M/w : Note that the moment in steady flight (M0) is zero. Further

M 1 M 1 = = S c u0 Cm w u0 2 Mw = 1 M Q Sc = Cm Iyy w u0 Iyy

(4.122) (4.123)

It may be recalled from chapter 2 Eq.(2.65), that:

Cm = CL (
Remark:

xcg c

xac c

) + (Cm )f , n, p

d VH CL t (1 ) d

(4.124)

It will be shown later that through this term the static stability influences the dynamic stability. 89

4.6.3. Derivatives due to change of q These derivatives include Z/q and M/q; X/q is negligible. It may be recalled from discussion in section 2.9 that when an airplane has a rate of pitch (q), the components of airplane experience a vertical velocity of magnitude q x l ; l being the longitudinal distance of the component from c.g.. The contribution to the derivative due to q is mainly from the horizontal tail. I) Z/q : Referring to Fig.4.6 the change in lift of horizontal tail due to rotation is given by:

Lt = C L t

1 1 2 2 t Qt St ; Qt = Vt = u0 2 2

(4.125)
90

Fig.4.6 Stability derivative due to rate of pitch (Adapted from Ref.1.1, chapter 3)
91

Now Therefore Let

Z = Lt = CL t

qlt Qt St u0

(4.126) (4.127)

lt Z = CL t Q St q u0 CZ = Z 1 2 u0 S 2 q lt St C Z = C L t u0 S 2u0 C Z = = (qc / 2u0 ) qc q C Z

Then

C Zq

2u0 qc

CL t

lt St lt St (4.128) = 2CL t VH ; VH = u0 S c S
92

Q lt St Sc QSc 1 Z Zq = = CL t = CL t VH m q u0 c S m mu0 Zq = C Zq QSc 2mu0

(4.129) (4.130)

II) M/q: Refering to Fig.4.6 the change in the moment due to tail on account of rotation is:
(Mcg )t = lt L t

(4.131)

Or

(C mcg )t = VH CL t Cm

qlt U0

(4.132)

2u0 C m (Cmq )t = = (qc / 2u0 ) qc q = 2U0 qc C L t lt lt (4.133) VH = 2CL t VH 93 u0 u0

To account for the contributions to Cmq , due to other components , Ref.1.1, chapter 3 suggests that the value of Cmq in Eq.(4.133) be multiplied by 1.1. This yields : lt (Cmq ) airplane = 2.2 CL tVH C

1 M c QSc = Cmq Mq = 2u0 IYY IYY q

(4.134)

4.6.4 Derivatives due to time rate of change of angle of attack . These derivatives include Z and .M . According to
w w

reference 4.3, chapter 8, when the results of flight test data on stability were compared with the calculated 94 values, some discrepancy was observed.

. Analysis showed that terms involving should be included in the equations for w and q. The reason for this dependence appears due to the following . The flow from the wing takes some time before it reaches the horizontal tail. Hence the downwash angle () at the tail is not the that corresponds to at time t, but that at time (t-t). The time lag (t) is roughly equal to the time taken by the flow to reach horizontal tail (lt/u0). The lag in the angle of attack at tail is given i by:

d d lt d d lt d lt t = t = = = dt dt u0 d dt u0 d u0

(4.135)

This lag in angle of attack would cause a change in lift of tail (Lt) given by:

Lt = CLt t Qt St

95

St Consequently, C = Lt = C Z L t t QS S St d lt = CL t d u0 S

(4.136) (4.137) (4.138)

Let

C Z =

C Z ( c / 2u0 ) 2u0 c CL t

2u0 C Z c
i

St d lt d u0 S

d = 2CL t VH d
Zw = 1 Z c QS = C Z m w 2u0 m u0

(4.139)

(4.140)
96

By a similar argument the change in the moment due to tail owing to the lag is:

Mcg = lt Lt = lt C L t t Qt St
Hence C mcg = CL t
C m = Cm

(4.141)

d lt VH d u0

Cm 2 u0 Cm = i (c / 2u0 ) c lt d = 2 CL t VH = (Cmq )tail c d

(4.142)

1 M c Q Sc Mi = = Cm w 2u0 u0 Iyy Iyy w

(4.143)
97

4.6.5 Derivatives due to control deflections (Ze and Me):

CZ e = CL e Cm e St d CLt 1 Z QS = ; Z e = = CZ e S de m e m dCLt

(4.144) (4.145)

1 M Q Sc = VH ; M e = = Cm e de Iyy e Iyy

Remark: A summary of the longitudinal stability derivatives is given in Table 4.5. Equations for the non-dimensional quantities involved in expressions for these derivatives are given in Table 4.6. The symbols used are given in Table 4.7 for ready reference.
98

Xu = Zu =

(C Du + 2C D )QS mu0 (C Lu + 2C L )QS mu0

( s 1 ) ( s 1 ) (s ) (m / s 2 )
1

(CD CL )QS Xw = mu0 Zw = C z Z = u0 Zw Z e = C Z e QS / m Mw = Cm c QSc 2u0 u0 Iyy

(s 1 )

(C L + C D )QS Zw = mu0 Z = u0 Zw Zq = C Zq Mu = C mu Mw = C m c QS 2u0 m

c QS / (u0 m) 2u0 (m / s2 ) (m / s2 )

(m / s) ( ( 1 ) m.s

(QSc ) u0 Iy (QSc ) u0 Iyy

(m1 ) (s 1) (s 2 )

1 ) m.s

M = u0 Mw QSC ) M e = Cm e ( Iyy

M = u0 Mw Mq = C mq (

( s 2 )

Table 4.5 Longitudinal stability derivatives (Adapted from Ref.1.1, chapter 3)

c )(QSc ) Iyy (s 1 ) 2u0

99

X-force derivatives

Z-force derivatives

Pitching moment derivatives

CXu =[CDu + 2CD ] + CTu

C Xa 2CL CL = CL Ae

M 12 CZu = CL 2CL 2 1 M1

Cmu =
Cm = CL w (

Cm M1 M 1

CZ = (CL + CD )
CZ = 2 CL t

0 q 0
e 0

d C = 2 C V lt d VH m L t H c d d
c

xac ) + (Cm ) f ,n , p c c d VH CL t (1 ) d

xcg

CZq = 2 CL t VH Cmq = 2 CL t VH lt
CZ e = CL e = St dCLt S d e

dCLt Cm e = VH d e

Table 4.6 Expressions for longitudinal stability coefficients (Adapted from Ref.1.1, chapter 3)
100

4.6.6 Lateral stability derivatives The small perturbation equations for lateral motion are given by Eqs.(4.88), (4.89) and (4.90). These equations involve the stability derivatives denoted by Yv , Yp, Yr, Lv , Lp, Lr, Nv , NP , Nr , Yr , La , Lr , Na and Nr. Instead of Yv , Lv and Nv , some times Y , L and N are used. It may be pointed out Yv= Y/u0, Lv = L/u0 and Nv=N /u0. The expressions for these derivatives in terms of nondimensional quantities are given in Table 4.8. The expressions for the non-dimensional quantities are given in table 4.9. The geometric and other parameters of the airplane are given in Table 4.7 for ready reference. Brief explanations regarding important 101 contribution to non- dimensional quantities are

A- Aspect ratio b Wing span CD- Reference drag coefficient CL- Reference lift coefficient CL Airplane lift curve slope CLw Wing lift curve slope CLt H.Tail lift curve slope CLv- V.Tail lift curve slope _ c Mean aerodynamic chord e Oswalds span efficiency factor lt- Distance from c.g. to tail aerodynamic centre lv- Distance from c.g. to vertical tail aerodynamic centre VH- Horizontal tail volume ratio Vv- Vertical tail volume ratio

M1-Flight Mach number S-Wing area St Horizontal tail area Sv- Vertical tail area Zv- Distance from centre of pressure of vertical tail to fuselage centerline - Wing dihedral angle Wing sweep angle d/d Change in downwash due to a change in angle of attack Efficiency factor of the horizontal tail v- Efficiency factor of vertical tail factor Taper ratio d / d Change in sidewash angle with a change in

Table 4.7 List of symbols for stability derivatives 102 (Adapted from Ref.1.1, Chapter 3)

given below.

Cy:
The major contribution to this quantities is from the vertical tail. From Eq.(3.13) we get: 1 (4.146) (Y )v.tail = Vv2 S CL v v 2 Noting that v=+, v= Vv2 / V2 and Cy= Y / V2S , Eq.(4.146) gives: Sv d (4.147) Cy = - v CL v (1+ ) S d It may be added that
Y = QSCy m

(4.148)

Remark: Reference 1.5b, chapter 7 mentions that wing and fuselage also have small contributions to 103 Cy (see also Appendix C).

Cn: From Eq.(3.21):

d Cn = (Cn )w + (Cn ) f ,n, p + VV v CLv (1+ ) d

(4.149)

Details regarding estimation of various terms in Eq.(4.149) are given in Section 3.2.3 . Note that
N = QSb C n I zz

(4.150)

Cl: From Eq.(3.34):

C 'l = (C 'l )w + (C 'l )f ,n, p + (C 'l )vt

Details regarding estimation of different terms in Eq.(4.151) are given in section 3.3.1 to 3.3.4. Note that:
L ' = QSb C 'l I xx

(4.151)

104

Cyp: The contribution to this quantity comes mainly from the vertical tail. Consider an airplane rolling with angular velocity p. A component of the airplane which is at a distance z from the X-axis would be subject to a linear velocity pz. This would induce a sideslip angle = pz/u0. The vertical tail when subjected to this sideslip would produce a side force. The vertical location of this side force would depend on zv , lv and the angle of attack of the airplane in the undisturbed flight; zv is the distance of the a.c. of vertical tail above of c.g. Reference 1.5b, chapter 8 gives the following expression:
105

Cyp (zv cos lv sin) Cyp (Cyp )v.tail =2 Cyv ; Cyv = b u0

Note that:
Yp = QS bC yp 2 mu0

(4.152) (4.153)

Remark: Cyp is generally small and neglected. Clp : Main contribution to this quantity is from the damping due to wing. From Eq.(3.52):
(C 'l )damp 2C L w p = u0 Sb
b /2

cy 2 dy
b /2

(4.154) (4.155)
106

Hence

4 C L w C 'l = C 'lp = pb Sb2 ( ) 2u0

cy 2 dy

Note that: Remarks: i) ii)

L'lp =

' QS b 2 C lp

2 I xx u 0

(4.156)

The horizontal tail and vertical tail also make minor contributions to Clp (see Appendix C). Reference 1.1, chapter 3 gives a formula for Clp in terms of CLw and geometrical parameters of the wing (see table 4.9). Cnp: The main contribution is from the wing in the form of adverse yaw. As explained in section 3.2.6:

(C n ) adverse yaw

C L pb = 8 2u 0

(4.157) 107

dCn CL = Hence Cnp = pb 8 d( ) 2u0 Remark:

(4.158)

Reference 1.5b gives a procedure which includes contributions due to vertical tail (see also Appendix C). Cyr: When an airplane has an angular velocity in yaw (r) , then a components of the airplane at a distance x from the c.g. will experience a sideword velocity of r x. This would result in side slip of (r x/u0) leading to side force and consequently Cyr. The major contribution to Cyr comes from the vertical tail. It is evident that the vertical tail will have a side ward 108 velocity of r lv and side slip angle of ( -r lv/u0).

This would produce a side force : Y =- CL Qv Sv (4.159)

lv r C L v ( ) Qv S v u0 lv r Sv Hence C y = = C L v ( ) v QS u0 S
Let

(4.160)

S v lv Cy C yr = , Then C yr = 2 C L v v rb S b ) ( 2u0 Sv , S
we get:

(4.161)

Noting that (C y ) v.tail C L v v

C yr = 2 (C y )v.tail

lv b

(4.162) 109

Note that Remark:

Yr =

QSbCyr 2mu0

(4.163)

Yr is generally small and in Eq.(4.88) it appears as (u0-Yr). Hence Yr is often neglected. Cnr: As mentioned above when an airplane has an angular velocity r the vertical tail experiences a sideslip angle = (-lvr / u0) . The side force thus created would also produce a yawing moment given by: N = CLv Qv Sv lv lv r (4.164) = C L v ( ) Qv S v lv u0 Or
lv r N Cn = = C L v ( ) vVv qSb u0
110 (4.165)

Cn Let Cnr = , rb ( ) 2u0

lv lv 2 Then Cnr = 2 CL v v Vv 2 (C y ) v.tail ( ) b b
QSb2 Cnr Note that: N = 2I xx u0

(4.166)

(4.167) (4.168)

Remark: When the airplane has a rate yaw, the two wing halves will experience different velocities (see explanation of Clr) and hence experience different drags. This would cause a small contribution from wing to Cnr (see appendix C). Clr: Let us consider an airplane experiencing a positive 111 rate of yaw i.e. right wing back. Then a section of

the right wing at a distance y from the x-axis would experience a velocity r y or relative wind of (u0-r y) . Consequently the dynamic pressure will be lower on the right wing. For the same reason the dynamic pressure will be higher on the left wing. The lifts on the two wing halves will be different, thus producing a positive rolling moment. Strip theory as explanied in section 3.3.6 can be used to get an approximate estimate of Cl and Clr. In addition the side force on the vertical tail also contributes to Clr. Table 4.9 gives an approximate formula as given in Ref.1.1,chapter 3 . Reference 1.5b gives an improved estimates. See also Appendix c. Note:
QSb2 C 'lr L 'r = 2I xx u0

(4.169)
112

Lateral control derivatives: The derivatives Cla and Cnr have been discussed in section 3.3.6 and 3.2.5 . The cross derivatives viz. Cna and Clr can be obtained from Ref. 1.1 and 1.8.

113

Y = Yp = Yr = Y a =

QSCy m QSCyp 2mu0 QSCyr 2mu0 QSCy a m

(m / s ) (m / s) (m / s ) (m / s )
2

N = Np =

QSbCn I zz QSb2Cnp 2I xxu0

(s ) (s 1 )

L ' = L 'p =

QSbC 'l I xx QSb2C 'lp 2I xxu0

(s 2 ) (s 1 )

QSb2Cnr 1 Nr = (s ) 2I xxu0 Y r = N r = QSCy r m (m / s 2 )

QSb2C 'lr 1 L 'r = (s ) 2I xxu0

N a =

QSbCn a 2 (s ) I zz

QSbCn r (s 2 ) I zz (s 2 )

QSbC 'l a 2 (s ) L ' a = I xx

QSbC 'l r L ' r = I xx

Table 4.8 Lateral stability derivatives (Adapted from Ref.1.1, chapter 3)

114

Y-force derivatives

Yawing moment derivatives

Rolling moment derivatives

Sv d C y = CL v (1 + ) S d

Cn : see Eq.(4.149) C 'l : see Eq.(4.151)

CL Cnp = 8
C L 1 + 3 C ' lp = 12 1 +

P Cyp: see Eq.(4.152) r a

lv C yr = 2 ( ) (C y )v.tail b

Cna:See Ref.1.1, C 'L a = Sb cy dy y ch.3 Clr:See Sv C y r = CL v r Cn r = Vv v CL v Ref.1.1,ch.3 S Table 4.9 Expression for lateral stability coefficients 0
1

l z C lv C 'lr = L 2 v v C y tail Cnr = 2 vVv ( )CL v 4 b b b y 2CL

2

(Adapted from Ref.1.1, chapter 3)

115

4.7 Examination of stability of longitudinal motion After deriving the linearized equation for small perturbation, it was mentioned at the end of section 4.5 , that the stability of the motion can be examined without obtaining the solution of the governing differential equations. The procedure is as follows. The small perturbation equations (Eqs.4.85 to 4.87) are linear. Such a set of equations admits a solution of the form: (4.170) u=1et, w= 2et, = 3et. Substituting these in the equations of longitudinal motion i.e. Eqs.(4.85),(4.86), (4.87) and noting that 116 for stability problem e and t are zero, we get:

1et Xu1et Xw2et+ g cos 0 3 et =0 . -Zu 1 et +[(1-Zw) Zw]2et [(u0+Zq) -g sin0] 3et=0) . . -Mu1et +(Mw +Mw)2 et + (2- Mq) 3 et=0 Dividing by et yields:

(4.171)

(4.172) (4.173)

( X u )1 Xw 2 + g cos 0 3 = 0
w

(4.174)

Zu 1 + {(1 Z i ) Zw }2 + {(u0 Zq ) g sin0}3 = 0 (4.175) Mu 1 + (M i + Mw )2 + (2 Mq)3 = 0

w

(4.176) 117

The Eqs. (4.174),(4.175), (4.176) are a set of homogeneous equations for1,2 and 3 . The solution 1=2=3=0 is called a trivial solution for obvious reasons. For a non-trivial solution to exist, we must have:

Xu
Zu Mu

Xw
w

g cos0
(4.177)

(1 Z i ) Zw (u0 Zq ) g sin0 = 0 M i + Mw
w

2 Mq

When the determinant in Eq.(4.177) is expanded, it yields the characteristic equation of the dynamic system : (4.178) A4+B3+C2+D+E = 0. When Zq and Zw are ignored and 0 is taken zero, the 118 coefficients A,B,C,D and E in Eq.(4.178) are :

A =1 B = M q u0 M i Z w X u
w

C = Zw Mq u0Mw Xw Zu + Xu (Mq + u0M i + Zw )

w

(4.179)

D = Xu (Zw Mq u0Mw ) + Zu (Xw Mq + gM i ) Mu (u0 Xw g)

w

E = g(ZuMw Zw Mu )

Equation (4.178) has four roots namely 1,2,3 and 4. The response to the disturbance i.e. the variations of u, w and with time can now be written as: (4.180) u= e 1 t + e 2 t + e 3 t + e 4 t
11 12 13 14

w = 21 e 1 t + 22 e 2 t + 23 e 3 t + 24 e 4 t = 31 e 1 t + 32 e 2 t + 33 e 3 t + 34 e 4 t

(4.181)
119 (4.182)

To evaluate the coefficients 11, 12,.. , 33 and 34 we would need to solve the differential Eqs.(4.85) to (4.87) with appropriate initial conditions. However to examine the stability we only need to know the values of 1 to 4, because the term et will ultimately decide whether the disturbances u, w and die down or not. The roots can be of the following six types: a) is real and positive =r b) is real and negative =-r c) is zero When the roots are complex they appear as a pair of complex conjugates (r+is) and (r-is) . Where r is the real part, s is the complex part and i is (-1) .The
120

two roots are together represented as (ris). There could be three cases when roots are complex. d) 1 and 2 constitute a complex pair r is with r positive. e)1 and 2 constitute a complex pair r is with r negative. f) 1 and 2 constitute a complex pair r is with r being zero. The variations of et with the time for these six cases are explained below and shown in Fig.4.7. I) When the root is real and positive the term et become ert and it is evident that the disturbance will grow exponentially (Fig. 4.7a). This response is called divergence. 121 II) When the root is real and negative the term et

(a) divergence

(b) subsidence

(c) Neutral stability

(d) Divergent
oscillation

(e) Damped oscillation

(f) Undamped oscillation

122

Fig.4.7 Motions following disturbance - as indicated by roots

becomes e-rt . This indicates that the disturbance will die down eventually (Fig.4.7b). This motion is called subsidence. III) If the root is zero the term et would become e0 or unity. This indicates that the system would remain in the disturbed position (Fig.4.7c). This response is called neutral stability. IV) When the roots form a complex pair, they appear as (r i s). There are two different possibilities. a) When the four roots consist of two real roots (1 and 2) and a complex pair (r i s), then the response would be of the form:
u=11 e1t + 12 e2t + '13 e-rt cos (s t + C1 )

(4.183) 123

where 13 and C1 are constants.

t t -rt -rt Or u=11 e + 12 e + 13 e cos (s t ) + 14 e sin (s t )
1 2

(4.184)

b) When the four roots consist of two complex pairs, (r1 i s1) and (r2 i s2), then the response would be of the form:

u= '11 e -r1 t cos( s1t + C 1 ) + '12 e -r2 t cos( s2 t + C 2 )(4.185)

u= 11 e-r1t cos(s1t ) + 12 e-r1t sin(s1t ) + 13 e-r2t cos(s2t ) Or

(4.186) As seen from Eqs.(4.183) and (4.185), the response 124 will be periodic when the roots are complex.

+ 14 e-r2t sin(s2t )

The amplitude of the oscillation is decided by the quantity r. Following three types of responses are possible depending on the value of r. i) If r is positive, then the amplitude of the periodic variation will increase with each oscillation (Fig.4.7d). This mode is called divergent oscillation. ii) If r is negative then the amplitude of the periodic variation will decrease with each oscillation (Fig.4.7e). This mode is called damped oscillation. iii) If r is zero, then the amplitude of the periodic motion remains constant (Fig.4.7f). This mode is called undamped oscillation. From the above discussion we observe that for a motion to be stable the root, when real should be 125

negative and when complex the real part should be negative. Thus we can conclude about the dynamic stability of the airplane by observing the roots of the characteristics equation. We do not need to obtain the response. To illustrate the aforesaid discussion we consider the dynamic stability of the general aviation airplane. Example 4.1 Examine the dynamic stability of the general aviation airplane whose details are given below. See Fig.4.8 for the three views of the aircraft . This example has been adapted from Ref.1.1, chapter 4. Flight condition: Steady level flight at sea level at u0= 53.65 ms-1 (M =0.158) weight of the airplane = 12251.4 N
126

Iyy = 4075.8 kg m2, Geometric details: _ 2, c= 1.74 m, b=10.18 m S=17.11 m Other details: = 1.225 kg m-3 , CL= 0.41, CD= 0.05 CL =4.44 rad-1 , CD = 0.33 rad-1, Cm = -0.683 rad-1, . CLu= 0 , CDu = 0 , Cmu = 0 , Cm = -4.36, CLq = 3.8 , Cmq= -9.96 . Solution: From the above data we deduce the following quantities needed for obtaining stability derivatives.
127

Fig.4.8 Three views of the general aviation airplane (Adapted from Ref.1.1, chapter 4)
128

1 1 2 Q = u0 = 1.25 (53.65)2 = 1763 Nm2 , QS = 30164 N, 2 2 c = 0.0162 s, mu0 = 67001.8 kg ms 1 QSc = 52486 Nm, 2u0 QS QSc = 0.450 , = 0.240 mu0 u0 Iyy

The longitudinal stable derivatives are:

Xu = QS (CDu + 2CD ) = 0.45(0 + 2 0.05) = 0.045 s 1 mu0

QS (CLu + 2CL ) = 0.45(0 + 2 0.41) = 0.361 s 1 Zu = mu0 Mu = 0 Xw = Zw = QS (CD CL ) = 0.45(0.33 0.41) = 0.0357 s 1 mu0 QS (CL + CD ) = 0.45(4.44 + 0.05) = 1.978 s 1 mu0
129

Mw =
w

QSc C m = 0.240 (0.683) = 0.164 m 1 s 1 u0 Iyy

Xi = 0 Zi = 0
w

M i = C m
w

c QSc = 4.36 0.0162 0.240 = 0.01695 m 1 2u0 u0 Iyy

Xq = 0 Zq = 0 Mq = C mq c QS = 9.96 0.0162 53.65 0.240 = 2.078 s 1 2u0 Iyy

Substituting the numerical values of the stability derivatives into Eqs.(4.179) we get: A=1, B=5.01, C=13.15, D=0.6684 and E=0.581. Hence the stability quartic ( Eq.4.178) or the characteristic equation appears as:
130

 4 + 5.01 3 + 13.15 2 + 0.6684 + 0.581 = 0

Solution of the characteristic equation:

(4.187)

The characteristic equation is a fourth degree polynomial. Exact solutions for polynomial equations are available only up to polynomials of degree three. For the present case we adopt an iterative solution technique. It may be added that the iterative technique to be used for the fourth degree polynomial depends on the magnitudes of the coefficients of the terms in the polynomial. The characteristic equation of the longitudinal motion is generally such that the coefficients of 3 and 2 are much larger than the coefficient the term and the constant term (see values of B,C,D and E in Eq.4.187). Following Ref.4.3 Appendix 4, the iterative procedure
131

for solving Eq.(4.187) can be described as follows: (4.188) f () = 4 + B3 + C2 + D + E Let Let us express f() as a product of two quadratics i.e.:
f () = (2 + b + c)(2 + + )

(4.189)

Expanding R.H.S. of Eq.(4.189) and comparing coefficients of terms in Eqs. (4.188) and (4.189) we get: b+ = B c + b + = C (4.190) b + c = D c = E For longitudinal motion of airplane , the coefficients D and E are generally much smaller than B and C and hence and in Eq.(4.190) are much smaller than b and c of the same equation. Hence, as a first 132 approximation:

b1 B c1 C E E 1 = c C cD bE CD BE 1 = 2 c C2

Hence

We work out the roots of the two quadratics in Eq.(4.191). For the second approximation, we take: b2 B 1 c2 C b1 1 1
c2 D b2 E 2 2 c2

CD BE E f () ( + B + C) ( + + )0 2 C C
2 2

(4.191)

2 =

E c2

133

We again work out the roots of the quadratics obtained using b2 , c2, 2 and 2 . The procedure is continued till the roots obtained in the two consecutive approximations do not change significantly in their value. Application of this procedure in the present case is described below. Here we have A=1, B=5.01, C=13.15, D=0.6684 and E=0.581. First approximation: b1 = 5.01; c1 = 13.15; 1 = (0.581/13.15) = 0.0442; 1=(13.15 x 0.6684 - 5.01x0.581) / 13.152 = 0.0340 Thus, f() (2+5.01+13.15)(2+0.034+0.0442) The roots of the two quadratics are:
134

1,2 =-2.505 i 2.622, 3,4 =-0.0170 i 0.2095 Second approximation : b2 = B - 1 = 5.01 0.034 = 4.976. Similarly, c2=12.94, 2 = 0.0344, 2 = 0.0449 The roots of the quadratic obtained using b2 , c2, 2 and 2 are : 1,2 =-2.488 i 2.598 , 3,4 =-0.0172 i 0.211 Carrying out the iteration once more, we get: 1,2 = -2.488 i 2.596, 3,4 =-0.01718 i 0.2113. The values of the roots do not seem to change significantly from the second to the third iteration and we can stop the process. Remarks: i) In section 4.9 we will express the equations of
135

motion in state space variable form and then the roots of the characteristics equation can be obtained by using packages like Matlab. ii) It is seen that the four roots of the characteristic equation consists of two pairs of complex roots. The real parts of both the roots are negative and the airplane has dynamic stability for the given flight conditions and configuration. Discussion on the modes of longitudinal motion is given in the next section. iii) Rouths Criteria: Presently we can obtain the roots of the stability quadratic using packages like Matlab. However earlier the tendency was to look for elegant analytical / approximate solutions. Rouths criteria is 136 a method which indicates whether the airplane would

be stable without solving the characteristic equation. We present the criteria without giving the proof. A quartic A4+B3+C2+D+E = 0 will have stable roots i.e. real roots negative and complex roots with negative real part when A>0 and the functions T1,T2,T3 and T4 are positive.
T1 = B T2 = B B 0 B T4 = 0 0 A D C = BC AD

A 0 B = DT2 B2E = BCD B2E AD2 D 0 A = ET3 B E

T3 = D C

(4.192)

A 0

D C

E D C 0 0 E

137

In the case of longitudinal stability quartic with A=1, the criteria simplify to: B>0 ; D>0; E>0 and R= T3 = BCD-B2E-AD2>0 The term R is called Rouths discriminant . iv) Damping and period of oscillation: A complex root is commonly written as : = i, When is negative the motion is damped oscillation. The damping is characterized by the time taken for the amplitude of the oscillation to become half and this time is denoted by t1/2. Since the amplitude of the oscillation is decided by et the time t1/2 is given 138 by: (4.193)

e
Or

t1/2

= 0.5
(4.194)

t1/2 = (ln 2) / || = 0.693 / ||

When is positive the motion is a divergent oscillation. The time to double the amplitude is a measure of the rate of divergence. This time can be denoted by t2. It is easy to show that: (4.195) t2 = (ln 2)/|| . When the root is real , its sign decides whether the motion would be divergence or subsidence. If the magnitude of the root is then the time to halve the amplitude or double the amplitude are given by Eq.(4.194). When the root is = i, the time period (P) is
139

given by: (4.196) P = 2/ The number of cycles for amplitude to halve (N1/2) is: (N1/2) = t1/2/P. (4.197) Applying the above formulae to the roots of the general aviation airplane, we get: a)1,2 = -2.436 i 2.521: t1/2=0.693/2.436= 0.284 s, P=2 /2.52 = 2.492 s, N1/2= 0.284 / 2.492=0.114cycles b) 3,4 =-0.0165 i 0.2144: t1/2=0.693/0.0165= 42 s P= 2 /0.2144 = 29.3 s, N1/2=42 / 29.3=1.43 cycles
140

4.8 Modes of longitudinal motion The motions represented by the different roots of the characteristic equation are called the corresponding modes. For the longitudinal motion the characteristic equation generally has two complex roots one of them has short period and heavy damping (see example 4.1). This mode is called short period oscillation (SPO). The other mode has long period and low damping (see example 4.1). Because of low damping it takes a long time to die. However the period is long and pilot has enough time to take corrective action. To appreciate further the features of these two modes we consider the response of the airplane to a disturbance. Reference 1.8, chapter 6 presents the 141

response of the general aviation airplane discussed in example 4.1. The disturbance consists of =50 and u=0.1u0 = 17.6 ft s -1 or 5.365 m s-1 Figures 4.9 a and b show the changes, with time, in the angle of attack() and pitch rate (q). Figures 4.10 a and b show the changes, with time, in the perturbation velocity (u) and the pitch angle() . Note that Fig.4.9 shows the initial response to disturbance i.e. upto t=20 s whereas Fig.4.10 shows the response upto t=200 s. It may also be pointed out that the solid curves in Figs. 4.9 and 4.10 represent the response of the airplane when the full set of equations, namely Eqs.(4.85), (4.86) and (4.87) is used. This set of equations is referred to as 142 fourth order system. The dotted lines represent

Fig.4.9a Response of general aviation airplane to disturbance in and u -change in (Adapted from Ref.1.8, chapter 6)
143

Fig.4.9b Response of general aviation airplane to disturbance in and u -change in q (Adapted from Ref.1.8, chapter 6)

144

Fig.4.10a Response of general aviation airplane to disturbance in and u -change in u 145 (Adapted from Ref.1.8, chapter 6)

Fig.4.10b Response of general aviation airplane to disturbance in and u -change in (Adapted from Ref.1.8, chapter 6)
146

approximate solutions to short period oscillation and phugoid motions. These approximations are considered in section 4.11. From Fig.4.9a it is seen that rapid changes in angle of attack and pitch rate take place in the first few seconds. Figure 4.10 show that the changes in velocity are negligible in the first few seconds and extend over a long period of time. Keeping these observations in mind and noting the values of the roots of the characteristics equation the following statements can be made. a) The short period oscillation with heavy damping influences the motion during the first few seconds. In this period the angle of attack and pitch rate 147 change rapidly whereas the velocity remains

approximately constant. However within this short time the angle of attack is nearly restored to its initial undisturbed value and remains so thereafter. b) The long period oscillation persists after the SPO has died down and influences the changes in velocity; the angle of attack remains almost constant. The total energy of the airplane is nearly constant during this motion and there is an exchange between the kinetic energy and potential energy of the plane. As the flight speed decreases the airplane loses kinetic energy and gains potential energy. It reaches a crest when the velocity is minimum (Fig 4.11). Conversely when the velocity increases the altitude decreases and the flight path shows a minimum when the velocity is maximum 148 (Fig.4.11). Figures 4.10a and b also show that after

Fig.4.11 Motion of airplane in LPO (Adapted from Ref.1.8, chapter 6)

149

the initial transient motion due to SPO dies out, the motion is governed by the LPO whose period of about 29 s. The amplitude of the velocity fluctuation is nearly half of the initial amplitude after about one and half cycles (note that N1/2 in example 4.1, is 1.43 cycles). However it takes long time for the motion to die out. Remark: The following website indicates phugoid motion as movement of horizon when photographed from inside the airplane: www.youtube.com . 4.9 Equations of motion in state space or state variable form The governing equations for the longitudinal motion (Eqs.4.85, 4.86 and 4.87) are ordinary differential equations with constant coefficients. When such 150 equations are written as a system of first order

differential equations, they are called state space or (4.198) state variable equations and are written as: i where X is the state vector, is the control vector and the matrices A and B contain stability derivatives. The steps for expressing the governing equations of longitudinal motions in state space variable form are as follows. The equations of motion are reproduced below for ready reference.
( d X u )u X w w + g cos 0 = X e e + X T T dt d d Zu u + [(1 Z i ) Zw ]w [(u0 + Zq ) g sin 0 ] w dt dt = Z e e + Z T T d d2 d M u u (M i + Mw )w + ( 2 Mq ) w dt dt dt = M e e + MT T

X = A. X + B.

151

. For simplicity, let us take 0 = 0 and zq = zw = 0. We observe that the third equation in the above set involves second derivatives of , whereas the state variable form of the equations has only the first derivative of the variables on the left hand side. To take care of this, we introduce a new variable q which is the derivative of . Then, the above set can be written as: (4.199) u = X u + X w + 0 q g + X + X
u w

( )

w = Zuu + Zww + u0q + (0) + Ze e + Zt T

w

(4.200)

q = Muu + Mi w + M w + Mq q + (0) + M e e + M t T (4.201) w = q

(4.202) We observed that the Eq.(4.201) in this set involves the 152 derivative of w on the right hand side, which is not in

accordance with the state variable representation . (Eq.4.198). To eliminate w , the derivative of w from Eq.(4.200) is substituted in Eq.(4.201). Consequently,

q = Mu u + M i ( Zu u + Zw w + u0 q + (0)
w

+ Z e e + Z t T ) + Mw w + Mq q + M e e + M t T or q = (Mu + M i Zu )u + (Mw + M i Zw )w + (Mq + M i u0 )q

w w w

+ (0) + (M e + M i Z e ) e + (M t + M i Z t ) T (4.203)
w w

The governing equations can now be written in matrix form as :

153

Xu u w Zu = q Mu + M i Zu w 0 X e + Ze M e + M i Z e M T w i.e., X = A.X + B.

Xw Zw Mw + M i Zw
w

0 g u 0 w u0 Mq + M i u0 0 q w 1 0

X T e ZT T + M i Z T w

(4.204)

u u w w i , = X = ,X = q q

e , T

(4.205)

154

Xu Xw Zu Zw A= Mu + M i Zu Mw + M i Zw Mq w w 0 0 X e X T Z e Z T B= M + M Z M T + M i Z T i e e w w

0 g u0 0 + M i u0 0 w 1 0

(4.206)

Remarks: i) The quantities u, w, q and are called state variables and e and t are called control variables. ii) When = 0, Eq.(4.205) reduces to: . (4.207) X=A. X. This set of equations has a solution: (4.208) X = X e r t Substituting Eq.(4.208) in Eq.(4.198) we get:
r

155

(r I A)Xr = 0

(4.209)

Where the identity matrix I in this case is:

1 0 I= 0 0 0 0 0 1 0 0 0 1 0 0 0 1

(4.210)

For a non-trivial solution to exist: |rI-A|=0.

(4.211)

Thus rs are the eigen values of matrix A. These are also the roots of the characteristics equation (Eq.4.178). Thus the roots of the characteristic equation can be obtained by finding out the eigen values of matrix A using packages like Mathematica, Matlab etc. 156

iii) For the general aviation airplane of example 4.1 the matrix A is :

9.81 0.0357 0 0.045 0.361 1.978 53.65 0 A= 0.00612 0.1305 2.9873 0 0 0 1 0

The roots of this equations using Matlab are: -2.488 i 2.596 and -0.01718 i 0.2113 . Which are same as those obtained by the iterative procedure.

157

4.10 Approximations to longitudinal flight The response of an airplane to disturbances as discussed in section 4.8 shows that the changes in or w take place in the first few moments of the disturbed motion. During this, the changes in u are negligible. In the subsequent motion, the angle of attack remains fairly constant and the velocity changes in a periodic manner. This change in velocity is accompanied by a change in the altitude of the airplane which implies a gradual exchange between the kinetic and potential energies. 4.10.1 Approximation to SPO: Thus to analyze SPO, we can make the simplifying assumption that during this phase of motion, u is 158 zero. Which results in the following simplifications:

a)In the set of governing equations (Eqs.4.85 to 4.87) the equation corresponding to X-force can be dropped. b) From the other two equations the terms multiplied by u can be ignored. . c) Writing as q we get the following set of equations.

d w Zw w u0 q = 0 dt dq (Mw + Mw Zw )w (Mq + Mw u0 ) q = 0 dt
Or

(4.212) (4.213)

i Zw u0 w w = i M w + M i Zw M q + M i u0 q w w q

(4.214)
159

Now w can be replaced by u0 .

u0 M 1 M 1 M Further M = = = = u0 M w I yy I yy (w / u0 ) I yy w

(4.215)

. Similarly, Z = u0Zw and M. = u0Mw . Substituting these, the simplified equations for SPO become: Z 1 u0 = Z q ) Mq + M i q ( M + M i u 0 The characteristic equation for this mode can be obtained from |I - A| = 0 or
160

(4.216)

Z u0 Z M M i u 0

1 =0

(4.217)

(Mq + M )

Expanding this, we get:

Z Z (Mq + M i + ) + (Mq M ) = 0 u0 u0
2

(4.218)

Solving Eq.(4.218) we get:

SPO

Z Z 2 1 = [{ M q + M i + } {( M q + M i + ) u0 u0 2
(4.219) 161

Z 4( M q M )}1 / 2 ] u0

Remark: Substituting in Eq.(4.219) the values of the stability derivatives for the general aviation airplane considered in example 4.1, we get : SPO = -2.48 i 2.61 whereas the exact roots for SPO are 2.488 i 2.596. 4.10.2 Approximation to LPO: Here, we assume that or w/u0 is small and changes occur only in u and . Changes in d/dt are also slow . Hence, we can ignore the moment equation. Retaining only u and in the remaining equations of motion, we get:
162

Or

d ( X u ) u + dt d Z uu u0 dt Xu u = Zu u0

g = 0 = 0 g u 0

(4.220) (4.221)

The characteristic equation for this mode can be obtained from |I - A| = 0 or

Xu
Zu u0
2

=0

Zu g Xu + =0 u0

(4.222) (4.223) 163

LPO

Zu g 1 / 2 1 2 } ] = [X u { X u 4 u0 2

Remarks: i) Substituting in Eq.(4.223) the values of the stability derivatives for the general aviation airplane considered in example 4.1, we get : LP0 = -0.023 i 0.26 as compared to the exact value of -0.0172 i 0.2113 ii) From Eq.(4.223) the damping of LPO is Xu. It may be recalled that

Hence damping of phugoid depends on CD. Thus a streamlined airplane will have lower damping. iii) The frequency of phugoid depends on
Z g 1 2 {X u 4 u }1 / 2 2 u0
164

2QSCD Xu = mu0

Since (Xu)2 is much smaller than (4Zu g/u0) the frequency(LPO) is roughly equal to (-Zug/u0)1/2 Recall that:

2QSCL 2QS 2mg 2 g Zu = = = mu0 mu0 QS u0

(4.224)

Hence, the frequency of phugoid is given approximately by:

LP 0

g 2 u0

(4.225)

Thus the period of phugoid is proportional to flight speed (u0) or the period will be longer at higher flight speeds.
165

4.11 Influence of stability derivatives on SPO and LPO From the discussion in the last two sections we draw the conclusions presented in Table 4.10.
Stability Item affected derivative Influence Increase of ( Mq+M. ) increases damping Increase of M or static stability increases the frequency Increasing Xu increases damping Increase of Zu increases the frequency

. Mq+M
M or Cm Xu or CD Zu or CL

Damping of SPO Frequency of SPO Damping of LPO Frequency of LPO

Table 4.10 Influence of stability derivatives on SPO and 166 LPO (adapted from Ref1.1, Chapter 4)

Remark: The term E in the characteristics equation (Eq.4.178) depends on Cm. When Cm is zero, E becomes zero leading to a zero root ( =0 ). This indicates neutral dynamic stability or when the aircraft has neutral static stability it also had neutral dynamic stability as it should be.

167

4.12 Stability diagrams: From the expressions for the stability derivatives it is seen that they depend on a) flight conditions (h, V or M) b) weight of the airplane (W) and c) c.g. location. Hence the roots of characteristic equation will depend on the choice of these parameters. A diagram, which shows the variation of dynamic stability or the roots of the characteristic equation with some chosen parameter, is called a stability diagram. These diagrams can be classified as follows. (a) One parameter stability diagram, (b) root locus 168 plot and (c) two parameter stability diagrams.

4.12.1 One parameter stability diagram In this case we choose a parameter of interest and obtain roots of the characteristics equation for various values of the chosen parameter . Figure 4.12a shows an example of one parameter stability diagram. The flight condition is such that the lift coefficient (CL) is equal to unity. The chosen parameter is denoted by and defined as: mc (4.226) C (x x ) =
2 2 S kB L NP cg

Where kB = radius of gyration =(Iyy / m)1/2 It may be noted that is proportional to static margin and hence -Cm. Consequently high value of 169 indicates higher static stability. When is zero the

airplane has neutral static stability. In the present case or Cm is varied and the roots of the characteristic equation are calculated. Since is plotted on the X-axis in Fig.4.12a, the following special way is adopted to plot the roots. The imaginary part of the complex roots viz. in the root ( i ) is plotted in the upper part of Y-axis. The real parts of the roots are also plotted on the Y-axis. The scales on the upper and lower parts of the y-axis may not be same. Let us now consider the points in figure 4.12 (a ) a) When is large, say 14, then the roots consist of two complex pairs namely -3.2 i 3.2and-0.06 i 0.47. It may be noted that the first pair represents SPO and the 170 second pair LPO. In the convention for plotting

Fig. 4.12a One parameter stability diagram CL=1 (Adapted from Ref.4.3, chapter 9)

171

adopted here ,the points A1 and B1 in Fig.4.12 a represent the SPO and SPO respectively. The points C1 and D1 represent LPO and LP0 of the second pair (LPO) respectively. b) As decreases to a value of nearly six , the roots are represented by points A2,B2, C2 and D2 . It is seen that the damping of the SPO does not change . However the period of SPO increases (SPO decreases). It may be recalled from Table 4.10 that the inference of Cm is to affect frequency of SPO. The damping of the LPO increases slightly and the period increases slightly. c) At around 3.2 the SPO breaks into two subsidence i.e. instead of a complex pair (SPO i SPO), there 172 are two equal real roots 1 and 2 with SPO =0 (see

Points A3 and B3).The damping and period of LPO change marginally. d) At around 1 there are two real roots (1 and 2 ) and a complex pair. The real roots are indicated by points A4 and B4. e) At 0.3 the LPO also breaks into two subsidences (3 and 4) as seen by points C5 and D5. f) As decreases further, four real negative roots are seen. When =0, the roots are represented by A6, B6 , C6 and D6. Note that D6 equals zero which represents neutral stability. This is due to the fact that when =0 , Cm is also zero and airplane has neutral static stability. g) For <0, one positive root is seen (see the dotted

173

line). When is less than about 0.4 a positive root, a complex pair and a large negative root are observed. Remarks: i) Generally low speed airplanes have enough static margin and the roots consist of two complex pairs. But this feature changes when the static margin is low. Fighter airplanes with negative static margin (relaxed static stability) will be unstable and need special controls for satisfactory flying.

ii) Figure 4.12 b shows the one parameter stability diagram at CL =0.25. As mentioned in section 4.10 and Table 4.10, the SPO is not much affected by CL, 174 but the period of phugoid (LPO) increases .

Fig. 4.12b One parameter stability diagram CL=0.25 (Adapted from Ref.4.3, chapter 9)

175

4.12.2 Root locus plot: This is another way of plotting the variation of the roots with a chosen parameters. Let the complex root be r i s . A root locus plot is obtained when r is plotted on the X-axis and s on the Y-axis. A root locus plot corresponding to the information presented in Fig.4.12 a, is plotted in Fig.4.13. Following is observed . a) When =14 , the roots are 1,2 = -3.2 i 3.2 and 3,4 = -0.06 i 0.47 . Points A1 and B1 in Fig.4.13 represent the roots 1 and 2 of SPO i.e. -3.2 + i 3.2, and -3.2 - i3.2 . Points C1 and D1 represents the roots 3 and 4 of LPO i.e. 0.06 + i 0.47 and -0.06 i 0.47. The value of is marked near the 176 points A1, B1, C1 and D1 in Fig.4.13.

Fig. 4.13 Root locus plot CL=1.0 (Adapted from Ref.4.3, chapter 9)
177

b) As decreases the damping of SPO is not affected but the period increase ( decreases). At 3.2 the SPO breaks into two subsidences. As the damping of SPO is not affected by , the variation of the roots of the SPO for >3.2 is shown by a vertical line in Fig.4.13. For <3.2 the roots are real and negative. Consequently points corresponding to these values of lie along the X-axis. c) As regards the LPO, the damping slightly increases as decreases . Further the imaginary part (LPO) slightly decreases as decreases . At 0.3 the LPO breaks into two subsidence. Thus the curve representing the root locus plot for the LPO is not a straight line but a curve as shown by the line PQR. 178

d) As decreases below 0.3, two real roots corresponding to LPO also lie on X-axis. When equals zero , a zero root is seen. e) When is negative , there is a positive root corresponding to LPO. At < - 0.4 the four roots are - one negative root, one complex pair and a large negative root. The damping of the oscillatory mode decreases with further decrease of . This behavior is shown by the curve KJL in Fig.4.13.

179

4.12.3. Two parameter stability diagram While discussing the stability criteria, it was pointed out that for the characteristic equation to have stable roots, the coefficient E in Eq.(4.178) and the Rouths discriminant R (Eq.4.193) should be positive. Thus lines corresponding to E=0 and R=0 would separate stable and unstable regions. Hence if we have a stability diagram in which we plot two parameters one on the X-axis and the other on the Y-axis, then we can plot these lines (E=0 and R=0) and they would serve as boundaries between stable and unstable regions. Following Ref 4.3, chapter 9, we choose and as two parameters. The quantity has already been defined (Eq.4.226) and is 180 proportion to Cm. The quantity is proportional to

St / S and is defined as:

Figure 4.14a shows E=0 and R=0 lines on - diagram for the LPO motion. When E<0 a negative real root or divergence is predicted. Hence a line corresponding to E=0 is referred to as divergence boundary. When R=0, the characteristic equation will have either a zero root or an imaginary root with real part as zero. A line corresponding to R=0 is called Routhian boundary. Figure 4.14a indicates the stable regions between the divergence boundary and Routhian boundary. It is seen further that for >0, it is not enough to have E>0 but

I yy 1 St CL t = ; iB = 2 2 S iB mlt

(4.227)

must have an appropriate value to ensure dynamic

stability.
181

Fig. 4.14a Two parameter stability diagram- LPO (Adapted from Ref.4.3, chapter 9)
182

4.13 Eigen values and eigen vectors As mentioned in section 4.9, a linear set of equations can be expressed as: . X=AX+B

(4.198)

Where X is the state vector and is the control vector. Further in stability analysis matrices A and B contain the stability derivatives. When is zero, the set reduces to: . (4.207) X=AX This set of equations has the solution. X=Xr e t
r

(4.208)
183

Where rs are the eigen values of the matrix A.

4.13.1. Eigen vector If j is an eigen values of a square matrix A, then a non-zero vector which satisfies the equation AX= j X Remarks: i) ii) Each eigen value has an associated eigen vector. Eigen values decide the nature of the motion following the disturbance and eigen vectors indicate the amplitude of the response. We follow the approach of Ref.1.8, chapter 5, to explain this with the help of an example.
184

(4.227a)

is called eigen vector corresponding to the eigen value j.

Let us consider a two degree of freedom system given by the following set of equations.

3x1 + 2 x1 + x2 = 0 x1 + 4 x2 + 3 x2 = 0

(4.227b) (4.227c)

By simple manipulation we can express this system in state variable form as:

x1 1 x = 2 11

3 x1 8 2 9 x 2

(4.227d)

The eigen values of the square matrix in Eq.(4.227 d) are given by:
185

| A I |= 0 , where I is the identity matrix

+
8 11 3 11 9 + 11 = 0

Or

2 11

expanding and simplifying gives 112 + 17 + 6=0 = -1, -6/11 To obtain the eigen vectors corresponding to these two eigen values we proceed as follows For =-1, using Eq.(4.227a) we get:
186

1 8 11 2

3 9

x1 x1 x = 1 x 2 2

(4.227e) (4.227f) (4.227g)

or -8x1+3x2=-11x1 and 2x1-9x2=-11x2 3x1+3x2=0 2x1+2x2=0

Simplifying Eqs.(4.227f & g) yields respectively:

Both of which are same . This means that the eigen vector is not unique and depends on the choice of one of the two variable. We choose x2=1, then x1=-1. Thus the eigen vector corresponds to =-1 is 187

 1 1 or [ 1

1 ]T

(4.227h)

where the superscript T indicates the transpose. Eigen vector corresponding to =-6/11: Using Eq.(4.227a) we get:

1 8 11 2

3 9

x1 6 x1 x = x 11 2 2

Or -8x1+3x2=-6x1 and 2x1-9x2=-6x2 Simplifying yields respectively : -2x1+3x2=0 2x1-3x2=0

188

Both of which are same . Choosing x2=1 gives x1=3/2. Hence the eigen vector corresponding to =-6/11 is [ 3/2 1 ]T. (4.227i) To clarify the physical significance of eigen values and vectors let us consider the solution as (recall section 4.7): (4.227j) t and x (t)= et x1(t)=1e 2 2 Substituting in governing equations (i.e. Eq.4.227b and 4.227 c) we get : 31+ 21+ 2=0 1+ 42+ 32=0 Which can be expressed as :
189

3 1 1 1 8 = (4.227k) 2 11 2 9 2 Comparing Eqs.(4.227d) and (4.227k) we note that the square matrix is same in both equations as it should be. Hence the eigen values of matrix in Eq.(4.227k) are -1 and -6/11 and the corresponding eigen vectors are [-1 1]T and [ 3/2 1]T respectively. This indicates that for =-1, 1=-1 and 2=1 . Then the first mode given by governing equations is :

x1=-e-t and x2= e-t With =-6/11, 1=3/2and 2=1 . Then the second mode given by governing equations is:
6 t 3 6 t x1 = e 11 and x2 = e 11 2
190

Thus the general solution is given by:

x1 = e t

6 3 11t + e 2 6 t 11

(4.227l) (4.227m)

x2 = e t + e

From Eqs.(4.227l and m) it is evident that the eigen values indicates the nature of the motion following the disturbance and eigen vectors indicate the amplitude of the response. Note: Actual responses of this system to a chosen disturbance and control input are presented in section 5.3.

191

4.13.2 Eigen vectors for the general aviation airplane Example 4.1 presents the stability derivatives and roots of the characteristic equation for the general aviation airplane flying at sea level at a flight velocity of 53.65 m/s. The roots are : short period oscillation (SPO): 2.488 2.596i; phugoid or long period oscillation (LPO): -0.01718 i 0.2113. To obtain the eigen vectors for these roots we follow the steps given in Ref.1.1, chapter 4. I) In the state space variable form the state variables in this case are u, w, q and . The governing 192 equation in matrix form are (Eq.4.204):

Xu u w Zu = q M u + M i Z u w 0

Xw Zw M w + M i Zw
w

0 u0 M q + M i u0
w

0 A

g u 0 w 0 q (4.204) 0

II) Following Ref.1.1, chapter 4, let us denote the elements of matrix A by A11, A12,A43, A44 , instead of Xu,Xw... 1, 0. Further let the root of the characteristic equation be denoted by j . The eigen vector corresponding to this root be denoted by [ uj wj qj j ]T . Then applying Eq.(2.227 a) to this case yields :
193

( j A11 )u j A12 w j A13 q j A14 j = 0 A21u j + ( j A22 )w j A23 q j A14 j = 0 A31u j A32 w j + ( j A33 )q j A34 j = 0 A41u j A42 w j A43q j + ( j A44 ) j = 0
In Eq.(4.227n) the stability derivatives A11,A12..A44 and the eigen value j are known . Only unknowns are the elements of eigen vector namely uj, wj, qj and j. As noted in the previous section, the eigen vector is not unique and out of its four elements, we should express three in terms of the fourth one. Let us express uj, wj and qj in terms of j. Dividing the first three equations of Eq.(4.227 n) by j and dropping the fourth equation 194 yields: (4.227n)

 u w q ( j A11 ) A12 A13 = A14 j j j u w q A21 + ( j A22 ) A23 = A24 j j j u w q A31 A32 + ( j A33 ) = A34 j j j

(4.227o)

The set of Eq.(4.227 o) when solved by standard techniques yields the eigen vector
u w q [ 1]T

The results are presented in table 4.11. It may be pointed out that (a) in table 4.11 the eigen vectors are in / / / (2 non-dimensional form i.e. u u , w u , [qc u )] (b) The elements of eigen vector are complex as the roots are 195 complex.
0 0 0

Eigen vector element

u / u0

Long period oscillation

=-0.01718 i 0.2113 -0.114 i 0.837

Short period oscillation

=-2.488 i 2.596 -0.034 i 0.025 1.0859 i 0.733

w / u0 0.008 i 0.05 = [qc / (2u0 )]

-0.000027 i 0.00347

-0.039 i 0.041

Table 4.11 Eigen vector for general aviation airplanelongitudinal motion (Adapted from Ref.1.1, chapter 4)
196

The elements of the eigen vectors give idea about relative magnitudes of the motion variables in the corresponding mode . When these ratios are complex numbers, as happens in the present case, they can be plotted in a vector or Argand diagram. Figure 4.14b and 4.14c present the information for LPO and SPO. From Fig.4.14b and Table 4.11 it is seen that the real parts of w / u0 (qc / 2u0 ) are very small and are not seen and in the diagram. The magnitude of
u / u0 is 0 .1 1 2 + 0 .8 3 7 2

or 0.8442. The phase of

u / u0

is given by tan-1(-0.837/0.11)=97.5o

197

Table 4.14b Argand diagram for eigen vectors of LPO (Adapted from Ref.1.1, chapter 4)
198

Table 4.14c Argand diagram for eigen vectors of SPO (Adapted from Ref.1.1, chapter 4)
199

From Fig.4.14 c and table 4.11 it is seen that for w / u0 SPO ( u / u0 ) and ( qc / 2u0 ) are very small. But ( )

1.08952 + 0.7332 = 1.313 has magnitude of 1 0.733 = 340 tan and phase of 1.0873 From the above discussion we can draw the following conclusions which were also drawn earlier while simplifying the equations for LPO and SPO in section 4.10. (i) The SPO is characterized by negligible changes in flight speed. The angle of attack oscillates with amplitude and phase not significantly different from . (ii) As regards LPO, the changes in q and are very small. u has significant magnitude and leads by 200 about 900.

4.14 Longitudinal stick free dynamic stability While defining the degrees of freedom we had mentioned that they are the number of coordinates needed to prescribe the position of any point on the system. A rigid airplane with controls fixed has six degrees of freedom viz. the three coordinates of the c.g. with respect to a ground fixed axis system and three Eulerian angles . When the elevator is free to rotate about its hinge, we have an additional degree of freedom i.e. to describe the position of a point on the elevator we also need to prescribe the elevator deflection (e); note the elevator is still assumed to be rigid. The rotation of the elevator causes changes in the 201

aerodynamic forces and the moments about the c.g.. . . . The third of the Noting that (Mcg)e = Mee +Me e governing equations (Eq.4.87) becomes :

d d2 d Mu u (M i + Mw )w + ( 2 Mq ) w dt dt dt M e e + M i e = 0
e
i

(4.228)

The new equation for the motion of elevator about its hinge would look like (Ref.1.4, chapter 10).
H H H H H H e = Ie ( + e ) u + + e + + + u e e + me xe (u0 uo + lt )

(4.229)
202

Where H= hinge moment, me= mass of the elevator, Ie = moment of inertia of the elevator and xe = distance of elevator c.g. behind hinge line. Thus the system of equation with elevator free will consists of four governing equations namely Eqs.(4.85), (4.86), (4.228) and (4.229). The characteristic equation for this set of equations would be of the form: (4.230) A6+B5+C4+D3+E2+F+G=0. where G depends on (dCm /dCL)free. It is found that the six roots of Eq.(4.230) form three complex pairs. Out of these three pairs the first two represent SPO and LPO which have been discussed 203 earlier. The third oscillatory mode represents the

oscillation of the elevator about its hinge. The period of this motion is about two seconds. This may sometimes lead to undesirable behavior as the response time of pilot is about one second. Hence the pilot action may reinforce the oscillations and cause instability. Proper damping of this mode is necessary. See section 10.6 of Ref.1.4 for further details. Self study topic: Handling characteristics of airplane . (These are discussed in Ref.1.1, at the end of chapter 4).

204

4.15 Lateral Stability The equations of motion for the lateral case (Eqs.4.88, 4.89,4.90) are reproduced below:
d ( Yv )v (u0 Yr )r g cos 0 = Y r r dt I d d L 'v v + ( L 'p )p [ xz + L 'r ]r dt I xx dt = L ' a a + L ' r r Nv v ( I xz d d + Np )p [ Nr ]r I zz dt dt

(4.88)

(4.89)

= N a a + N r r 1 L ' 1 N L 'v = ; Nv = ; I xx v I zz v

(4.90)

205

These equations can be put in the state variable form through the following steps (Ref.1.1, chapter 5):
v = Yv v + Ypp (u0 Yr )r + g cos 0 + Y r r

(4.231)

Ixz d p = L 'v v + L 'p p + r + L 'r r + L ' a a + L ' r r (4.232) Ixx dt Ixz d r = Nv v + Npp + p + Nr r + N a a + N r r Izz dt = p

(4.233) (4.234)

To eliminate d(r) /dt on the right hand side of Eq.(4.232), we substitute it using Eq.(4.233). The resulting equation would have the following term on 2 I xz d the right hand side: 206 p I xx I zz dt

Taking it to the left hand side yields:

2 Ixz d Ixz Ixz (1 ) p = (L 'v + Nv )v + (L 'p + Np )p Ixx Izz dt Ixx Ixx

Ixz Ixz Ixz +(L 'r + Nr )r + (L 'a + Na )a + (L 'r + Nr )r (4.235) Ixx Ixx Ixx
We introduce the following notations

L 'v Nv * , Nv = , etc. L = 2 2 I xz I xz (1 ) (1 ) I xx I zz I xx I zz
* v

(4.236)

207

Using the new notation given in Eq.(4.236), we have:

I xz * I xz * * p = (L + Nv )v + (Lp + Np )p I xx I xx
* v

+(L* + r

I xz * I I Nr )r + (L*a + xz N*a ) a + (L*r + xz N*r ) r (4.237) I xx I xx I xx

. Similarly eliminating p from r.h.s of Eq. (4.233) and . simplifying gives the following expression for r : I xz * I xz * * * r = (Nv + Lv )v + (Np + Lp )p I zz I zz
I xz * I xz * I xz * * * +(N + Lr )r + (N a + L a ) a + (N r + L r ) r (4.238) I zz I zz Izz
* r

208

The final form of the equations for lateral motion in the state variable form is:
Yv v * I * Lv + xz Nv I xx p = I * r Nv + xz L* v I zz 0 0 * I xz * N + L a + I xx a I N*a + xz L*a I zz Yp L* + p
* p

u0 Yr L* + r
* r

I xz * Np I xx

I xz * Nr I xx

I xz * N + Lp I zz 1

I xz * N + Lr I zz 0

g cos 0 v 0 p r 0 0

Y r I xz * a * N L r + I xx r r I xz * * N r + L r I zz

(4.239)

209

This can be put in the form:

X = A.X + B.

(4.240)

The solution of this equation, as in the longitudinal case, leads to,

| r I A |= 0 A1 4 + B1 3 + C1 2 + D1 + E1 = 0

(4.241)

4.15.1 Solution of lateral stability quartic The characteristic equation (Eq.4.241) is called lateral stability quartic. Reference 4.3, part III chapter 2 gives the following quartic and an approximate method to solve it. 210 f () = 0.968 4 + 5.723 + 20.32 + 97.2 + 21.3 = 0 (4.242)

A different iterative solution technique is used for solving this equation. It is seen that the coefficient E is much smaller than D. Hence a small root is expected. When the root is small, the terms with powers of would be very small in Eq.(4.242) and the first approximation(1(1)) can be written as: 1(1) = -E/D = -21.3/97.2 = -0.22 To get the second approximation we write Eq.(2.242) as: (+0.22)(0.9683+5.512+19+93) + 0.84 = 0. If 1(1) were the exact root , then the remainder should be zero; in this case, it is 0.84. We now get the second approximation as : 1
(2)

= -21.3/93 = -0.229.

211

Substituting in Eq.(2.242) , we get : (+0.229)(0.9683+5.52+19+92.9)+0.026=0 (4.243) Since the remainder now is very small, we take 1=-0.229. To get the other three roots, we consider the cubic part of Eq.(4.243) i.e. 0.9683+5.52+19+92.9 = 0 Dividing by 0.968 we get: 3+5.692+19.6+96 = 0. (4.244) To solve the cubic equation Eq.(4.244) we assume that the second root (2) is large. Then for the first approximation (2(1)), the second and third term in Eq.(4.244) can be ignored i.e. 3+5.692 =0 or 2
(1)

=-5.69.

212

to get the second approximation we substitute 2(1) in the neglected terms. Which gives:

(2) 2

19.6 96 =-5.69 = -5.22. 2 (-5.69) (-5.69)

In a similar fashion, we get, 2(3)=-5.77, 2(4)=-5.4, 2(5)= -5.37. The last two values are fairly close and we take 2 = -5.37. Substituting for 1 and 2 we get: f() = ( +0.229) ( +5.37) (2+0.32+17.3) = 0. the roots are -0.229, -5.37, -0.16 i 4.16 .
213

Remarks: i) The general aviation airplane discussed in example 4.1 has the following lateral stability derivatives (Ref.1.1, chapter 5). Yv= -0.254 s-1,Y = -13.94 m s-2, Yp=0, Yr=0, Lv =-0.298 m-1 s-1, L = -16.02 s-2, Lp = -8.4 s-1, Lr = 2.19 s-1, Nv = 0.082 m-1s-1, N = 4.49 s-2 , Np= -0.35 s-1, Nr = -0.76 s-1 , Ixx = 1423.4 kg m2 , Izz = 4795. 9 kg m2 , Ixz = 0. The lateral stability quartic is (Ref.1.1, chapter 5):

4 + 9.417 3 + 13.982 2 + 48.102 + 0.4205 = 0

(4.245)
214

The root in this case are : -0.00877, -8.435 and -0.487 i 2.335 . ii) In the above analysis we have included as a variable since it is directly involved in Eq.(4.88). We could have also included , the angular displacement in yaw . In this case we would get an additional governing equation namely, . (4.246) =r Reference 1.8 (chapter 6) and Ref.4.3 (chapter 7 part II) follow this approach. In this case Eq.(4.239) would get modified as:

215

Yv Yp u0 Yr v L* + Ixz N* L* + Ixz N* L* + Ixz N* p v Ixx v p Ixx p r Ixx r = I I I * * r Nv + xz L* Np + xz L* Nr* + xz L* Izz v Izz p Izz r 0 1 0 0 0 1 0 Yr * Ixz * * Ixz * a Na Lr + Nr + La + Ixx Ixx r * I * Ixz * * xz Na + La Nr + Lr Izz Izz

g cos0 0 0 0 0

0 v 0 p r 0 0 0

(4.247)

216

The characteristic equation corresponding to this set of equations would be :

A1 5 + B1 4 + C1 3 + D1 2 + E1 = 0

(4.248)

This equation has =0 as the fifth root in addition to the four roots discussed earlier. Similar result (i.e. a fifth degree polynomial as characteristics equation) would have been obtained if we solved Eqs.(4.88),(4.89) and (4.90) by assuming that: v=1et, =2et, =3et. (4.249)

217

4.15.2. Analysis of roots: The characteristic equation for the lateral motion has five roots. The motions indicated by these can be briefly described as follows I) equal to zero root : From Eq.(4.249) we observe that when is zero the solution would be: v=1, =2, =3. (4.250) Substituting these in Eqs.(4.88),(4.89) and (4.90) shows that v=0 but and need not be zero. Thus zero root , which means neutral stability , could result in airplane attaining constant angle of bank and / or yaw. Figure 4.15 shows a change in flight direction as the result of the zero root. See also 218 Figs.4.17a to 4.17d and discussion on disturbance to

Fig. 4.15 Directional divergence and spiral divergence (Adapted from Ref.1.1, chapter 5)
219

general aviation airplane. These constant values of and would need to be corrected with appropriate control action. II) Large negative root: It is called roll subsidence. It is heavily damped and does not pose any problem. III) Small negative root: The lightly damped real root is called spiral mode. When the root is positive i.e. unstable, the airplane looses altitude, gains speed, banks more and more with increasing turn rate. The flight path is a slowly tightening spiral motion (Fig.4.15). The difference between spiral motion and spin is that the angle of attack is below stall and the control surfaces are effective. IV) The complex root indicating oscillatory motion: 220 The motion corresponding to this root is called dutch

roll. It can be briefly described as follows a) b) c) d) e) Let the disturbance cause roll to left. Due to adverse yaw, this roll results in yaw to right, i.e. sideslip to the right. Because of the dihedral effect, the sideslip causes a restoring moment causing roll to right. Subsequently airplane yaws to left and rolls to left. This sequence of events continues and when the real part of the root is negative the amplitude of the oscillatory motion decreases with time. During this oscillatory motion the airplane is all along moving forward and its motion, according to Ref.1.1 (chapter 5) reminded someone of the 221 weaving motion of a Dutch ice skater. Hence the

f)

name Dutch roll (See Fig.4.16). Remark: Dutch roll is an undesirable mode as it causes discomfort to the passengers and may lead to missing of targets in military airplanes. It should have adequate damping. 4.15.3. Lateral-directional response of general aviation airplane Reference 1.8 chapter 6 presents the response of general aviation airplane discussed in example 4.1. The stability quartic for this airplane is given by Eq.(4.245) and the five roots are: =0, -0.00877, -8.435 and -0.487 i 2.335. It may be noted that:
222

Fig. 4.16 Dutch roll (Adapted from Ref.1.1, chapter 5)

223

a) The time to half amplitude (t1/2) of spiral mode is: t1/2= 0.693 =79 s. b) The period of the Dutch roll is : 2 / 2.335 = 2.69 s c) The time to half amplitude for the dutch roll is: 0.693 / 0.487 =1.42 s d) Number of cycles for half amplitude of Dutch roll is: 1.42 / 2.69 = 0.53 cycles The response to various disturbances are briefly discussed below. a) Disturbance in sideslip: Figure 4.17a shows the response to a disturbance of =50 at time t=0. It is seen that in the response to the Dutch roll , the 224

various modes show oscillatory motions which decay in two to three cycles. However the angle of yaw () does not go to zero but assumes a small value (Fig.4.17a). This is due to the zero root. b) Figure 4.17b shows the response to disturbance in bank angle =50 at t=0. The Dutch roll mode causes initial oscillations which decay rapidly. However the spiral mode has a low damping and it takes long time to decay. It is again noticed that the yaw angle () attains a steady state value due to one of the roots being zero. c) Figure 4.17c shows the response to disturbance in rate of roll, p=0.2 rad s-1 (12.5 deg s-1) at t=0. It is seen that the roll rate goes to zero without oscillation. 225

However sideslip, bank angle and the rate of yaw display oscillatory motions. The angle of yaw gradually attains a non-zero value again due to the presence of zero root. d) Figure 4.17d shows response to a disturbance in rate of yaw of r=0.2 rad s-1 (12.5 deg s-1) at t=0. It is observed that the response is similar to that in the other cases. Remarks: i) Since the governing equations are linear, the response to a combination of small initial disturbances can be obtained by summing up the responses given above. ii) For a video clip on motion of Basset airplane during 226 Dutch roll , see www.youtube.com.

Fig. 4.17a Lateral-directional response of the general aviation airplane- disturbance in (Adapted from Ref.1.8, chapter 6)
227

Fig. 4.17b Lateral-directional response of the general aviation airplane- disturbance in (Adapted from Ref.1.8, chapter 6)
228

Fig. 4.17c Lateral-directional response of the general aviation airplane- disturbance in p (Adapted from Ref.1.8, chapter 6)
229

Fig. 4.17d Lateral-directional response of the general aviation airplane- disturbance in r (Adapted from Ref.1.8, chapter 6)
230

4.15.4 Approximate analysis Reference 1.1, chapter 5 and Ref.1.8,chapter 6 present approximate analysis of spiral and Dutch roll modes. The important conclusions of the approximate analysis are as follows. a) The spiral root is approximately equal to:
L ' Nr L 'r N L '

spiral

(4.251)

Noting that L is proportional to dihedral effect and is negative, for spiral to be negative, the numerator in Eq.(4.251) should be positive or LNr-LrN > 0 or (L/N) > (Lr/Nr). b) The imaginary part of Dutch roll mode is approximately given as :
231

Dutch (N)1/2

(4.252)

Remark: The results of approximate analysis of lateral motion are not as good as those in the case of longitudinal motion. Hence it is recommended that exact roots be obtained in the lateral case. 4.15.5 Two parameter stability diagram for lateral motion: The area of the vertical tail (Sv) and the dihedral angle () are the two parameters which can be controlled by the designer to achieve desired level of lateral stability. The parameter Sv mainly affects N and affects L. Hence these parameters are suited for a two parameter stability diagram. Following 232 Ref.4.3, chapter 7, we choose the following

parameters: nv = Nv / ( u0 Sb) and lv= Lv /( u0 Sb) (4.252) Following observations can be made. I) Fig.4.18a shows a two parameter stability diagram for the lateral motion when CL= 0.1 which may indicate higher end of the flight velocity range. It may be noted that: a) The parameter 2 in the figure is defined as: 2 = 2m / ( Sb) (4.253) b) The spiral boundary denotes the case when E1 in the characteristic equation for lateral motion (Eq.4. 241) is zero. This is like the divergence boundary in longitudinal case (Fig.4.14). c) The Routhian boundary is obtained when the
233

Fig. 4.18a Two parameter stability diagram for lateral motion- CL=0.1 (Adapted from Ref.4.3, part-II , chapter 7)
234

Routhian (R) is zero. In the lateral case this boundary depends on 2. From Eq.(4.253) it is seen that 2 increases as decreases or the altitude increases. In Fig.4.18a it is seen that the Routhian boundary shifts closer to the spiral boundary as increases. II) Figure 4.18b shows the two parameter stability diagram for lateral motion when CL=1.0, which may indicate the low speed end of the flight velocity range. It is interesting to note that for 2 = 40 and 100, major part of the spiral boundary lies above the Routhian boundary. This indicates that either the spiral mode or the Dutch roll mode would be unstable. Above results indicate the need for investigating the 235 stability of the airplane under various combinations

Fig. 4.18b Two parameter stability diagram for lateral motion- CL=1.0 (Adapted from Ref.4.3, part-II , chapter 7)
236

of flight velocity , altitude and weight . Earlier practice , in situations where one of the roots is unstable, was to choose Sv and such that Dutch roll mode is stable as it is an undesirable mode . The present method is to use automatic control (see section 5.2 for a brief discussions on how automatic control alters level of stability). General Remark: See Appendix C for evaluation of the stability derivatives and the solution of stability quartic for Boeing -747 airplane.

237

Exercises 4.1 Answer the following. a) Explain the term degrees of freedom. How many degrees of freedom does a rigid airplane with free aileron, elevator and rudder have? b) Sketch a single parameter stability diagram or a root locus plot when Cm changes from a large negative value to a small positive value. c) The roots of a longitudinal stability quartic are: - 2.57 i 2.63 ; +0.02 and -0.26. Discuss the types of motions indicated by each mode. What would the final motion be?
238

4.2 Distinguish between the following a) Wind axes system and stability axes system. b) Short period and long period modes of oscillation of longitudinal motion. c) Spiral mode and Dutch roll mode of lateral motion. 4.3 The simplified equations for short period mode of longitudinal motion are given as:
d (w) Zw w u0 q = Z e dt d d (M w + Mw ) w + ( Mq )q = M e dt dt

239

a) write these equations as state variable equations. b) obtain the characteristics equation and its roots for the disturbed motion, given that: u0= 53.65 m /s , Zw = -1.978 s-1,Mw=-0.1602m-1s-1 , . Mw = -0.0167 m-1, Mq = -2.006 s-1. [Ans: Characteristic equation: 2 + 4.873 + 12.563 =0 = -2.437 i 2.574 ] 4.4 Briefly describe the Dutch roll mode of the lateral motion when an airplane is disturbed. The approximate form of this motion can be described in the state- space form as:
240

Y Yr Y r (1 ) u0 = u0 + u0 r r r N Nr N r

Given that Y=-2.6 m s-2 ; Nr = -0.34 s-1, Yr = -1.572 m s-2 , Yr = 0.741 ms-1, u0 = 51.33 ms-1; Nr= 0.616 s-2 , N = 0.64 s-2. a ) obtain the roots of the characteristics equation and examine the stability of motion to a disturbance. b) obtain the period of oscillatory mode and the time to damp to half amplitude. How will these values change if a yaw damper makes Nr to -0.68 s-1 ? [Ans: (a) without yaw damper: Dutch= -0.1955 i 0.781; t1/2 =3.54 s , period = 8.04 s.

241

(b) with yaw damper: Dutch = -0.3655 i 0.729 , t1/2 = 1.90 s , period = 8.61 s ]. 4.5 The longitudinal stability quartic of an airplane is: 4 + 43+ 102+ +3.8 =0 . Extract the roots and describe the motion depicted by them. (Adapted from Ref.4.3, part I, chapter 9). [Ans: 1,2 =-2.0287 i 2.394 ; 3,4 = 0.02873 i 0.6205 ].
242

4.6 The lateral stability quartic for an airplane is 4 + 163 + 13.1 2 + 9.8 + 0.73 = 0 . Extract the roots of this quartic. Obtain the time to double or halve the amplitude and the period of the oscillatory mode (s) (adapted from Ref.4.3, part II, chapter 7). [Ans: 1 = -0.0819 , t1/2 = 8.42 s ; 2 = -15.18 , t1/2 = 0.045 s; 3,4 = -0.369 i 0.667 , t1/2 = 1.86 s , period = 9.41s ]. 4.7 A system with two degrees of freedom has the following dynamical equations:
243

4 x + 2 x + 6 x + y + y 4y = 0 x + 2 x + 3x + 5y + 4y + ky = 0

where k is a constant. Obtain the characteristic equation and determine the value of k in order that the system may have neutral stability (i.e. one of the roots as zero). When k has this value, the characteristics equation has another root 2 = -1. Extract the remaining two roots and comment on the motion. (Adapted from Ref.4.3, part II, chapter 9). ( Remark: The characteristic equation corresponding to the above equations is a quartic. One root of the quartic is zero by choice. The cubic equation 244 governing the other three roots would be :

193+232+29+25=0 A hint has been given that one root of this cubic is -1. However this could have been guessed from the following property of polynomials. When the sums of the coefficient of the alternate terms are same the polynomial has =-1 as a root. In the above cubic we observe: 19+29 = 23+25). [Ans: k=-2 ; 1=0 , 2=-1 , 3,4 = -0.0105 i 1.142 ].

245

References 4.1 Miele, A. Flight mechanics, Addition-Wesley/ Pergamon Press, London, 1962. 4.2 McRuer, D., Ashkenas, I. and Graham,D. Aircraft dynamics and automatic control Princeton University Press, Princeton, N.J. 1973. 4.3 Babister A.W Aircraft stability and control Pergamon press, Oxford, 1961.

246