Flight Dynamics Summary

Flight Dynamics Summary 1. Introduction In this summary we examine the flight dynamics of aircraft. But before we do that, we must examine some basic ideas necessary to explore the secrets of flight dynamics. 1.1 1.1.1 Basic concepts Controlling an airplane To control an aircraft, control surfaces are generally used. Examples are elevators, flaps and spoilers. When dealing with control surfaces, we can make a distinction between primary and secondary flight control surfaces. When primary control surfaces fail, the whole aircraft becomes uncontrollable. (Examples are elevators, ailerons and rudders.) However, when secondary control surfaces fail, the aircraft is just a bit harder to control. (Examples are flaps and trim tabs.) The whole system that is necessary to control the aircraft is called the control system. When a control system provides direct feedback to the pilot, it is called a reversible system. (For example, when using a mechanical control system, the pilot feels forces on his stick.) If there is no direct feedback, then we have an irreversible system. (An example is a fly-by-wire system.) 1.1.2 Making assumptions In this summary, we want to describe the flight dynamics with equations. This is, however, very difficult. To simplify it a bit, we have to make some simplifying assumptions. We assume that . . . • There is a flat Earth. (The Earth’s curvature is zero.) • There is a non-rotating Earth. (No Coriolis accelerations and such are present.) • The aircraft has constant mass. • The aircraft is a rigid body. • The aircraft is symmetric. • There are no rotating masses, like turbines. (Gyroscopic effects can be ignored.) • There is constant wind. (So we ignore turbulence and gusts.) 1.2 1.2.1 Reference frames Reference frame types To describe the position and behavior of an aircraft, we need a reference frame (RF). There are several reference frames. Which one is most convenient to use depends on the circumstances. We will examine a few. 1 • First let’s examine the inertial reference frame FI . It is a right-handed orthogonal system. Its origin A is the center of the Earth. The ZI axis points North. The XI axis points towards the vernal equinox. The YI axis is perpendicular to both the axes. Its direction can be determined using the right-hand rule. • In the (normal) Earth-fixed reference frame FE , the origin O is at an arbitrary location on the ground. The ZE axis points towards the ground. (It is perpendicular to it.) The XE axis is directed North. The YE axis can again be determined using the right-hand rule. • The body-fixed reference frame Fb is often used when dealing with aircraft. The origin of the reference frame is the center of gravity (CG) of the aircraft. The Xb axis lies in the symmetry plane of the aircraft and points forward. The Zb axis also lies in the symmetry plane, but points downwards. (It is perpendicular to the Xb axis.) The Yb axis can again be determined using the right-hand rule. • The stability reference frame FS is similar to the body-fixed reference frame Fb . It is rotated by an angle αa about the Yb axis. To find this αa , we must examine the relative wind vector Va . We can project this vector onto the plane of symmetry of the aircraft. This projection is then the direction of the XS axis. (The ZS axis still lies in the plane of symmetry. Also, the YS axis is still equal to the Yb axis.) So, the relative wind vector lies in the XS YS plane. This reference frame is particularly useful when analyzing flight dynamics. • The aerodynamic (air-path) reference frame Fa is similar to the stability reference frame FS . It is rotated by an angle βa about the ZS axis. This is done, such that the Xa axis points in the direction of the relative wind vector Va . (So the Xa axis generally does not lie in the symmetry plane anymore.) The Za axis is still equation to the ZS axis. The Ya axis can now be found using the right-hand rule. • Finally, there is the vehicle reference frame Fr . Contrary to the other systems, this is a lefthanded system. Its origin is a fixed point on the aircraft. The Xr axis points to the rear of the aircraft. The Yr axis points to the left. Finally, the Zr axis can be found using the left-hand rule. (It points upward.) This system is often used by the aircraft manufacturer, to denote the position of parts within the aircraft. 1.2.2 Changing between reference frames We’ve got a lot of reference frames. It would be convenient if we could switch from one coordinate system to another. To do this, we need to rotate reference frame 1, until we wind up with reference frame 2. (We don’t consider the translation of reference frames here.) When rotating reference frames, Euler angles φ come in handy. The Euler angles φx , φy and φz denote rotations about the X axis, Y axis and Z axis, respectively. We can go from one reference frame to any other reference frame, using at most three Euler angles. An example transformation is φx → φy → φz . In this transformation, we first rotate about the X axis, followed by a transformation about the Y axis and the Z axis, respectively. The order of these rotations is very important. Changing the order will give an entirely different final result. 1.2.3 Transformation matrices An Euler angle can be represented by a transformation matrix T. To see how this works, we consider a vector x1 in reference frame 1. The matrix T21 now calculates the coordinates of the same vector x2 in reference frame 2, according to x2 = T21 x1 . 2 Let’s suppose we’re only rotating about the X axis. In this case, the transformation matrix T21 is quite simple. In fact, it is   1 0 0   T21 = 0 cos φx (1.2.1) sin φx  . 0 − sin φx cos φx Similarly, we can rotate about respectively,  cos φy  T21 =  0 sin φy the Y axis and the Z axis. In this case, the transformation matrices are,  0 − sin φy  1 0  0 cos φy and T21  cos φz  = − sin φz 0 sin φz cos φz 0  0  0 . 1 (1.2.2) A sequence of rotations (like φx → φy → φz ) is now denoted by a sequence of matrix multiplications T41 = T43 T32 T21 . In this way, a single transformation matrix for the whole sequence can be obtained. Transformation matrices have interesting properties. They only rotate points. They don’t deform them. For this reason, the matrix columns are orthogonal. And, because the space is not stretched out either, these columns must also have length 1. A transformation matrix is thus orthogonal. This implies that T T−1 21 = T21 = T12 . 1.2.4 (1.2.3) Transformation examples Now let’s consider some actual transformations. Let’s start at the body-fixed reference frame Fb . If we rotate this frame by an angle αa about the Y axis, we find the stability reference frame FS . If we then rotate it by an angle βa about the Z axis, we get the aerodynamic reference frame Fa . So we can find that      cos αa 0 sin αa cos βa sin βa 0 cos βa sin βa 0      (1.2.4) xa = − sin βa cos βa 0 xS = − sin βa cos βa 0  0 1 0  xb . 0 0 1 − sin αa 0 cos αa 0 0 1 By working things out, we can thus find that  cos βa cos αa  Tab = − sin βa cos αa − sin αa sin βa cos βa 0  cos βa sin αa  − sin βa sin αa  . cos αa (1.2.5) We can make a similar transformation between the Earth-fixed reference frame FE and the body-fixed reference frame Fb . To do this, we first have to rotate over the yaw angle ψ about the Z axis. We then rotate over the pitch angle θ about the Y axis. Finally, we rotate over the roll angle ϕ about the X axis. If we work things out, we can find that   cos θ cos ψ cos θ sin ψ − sin θ   (1.2.6) TbE = sin ϕ sin θ cos ψ − cos ϕ sin ψ sin ϕ sin θ sin ψ + cos ϕ cos ψ sin ϕ cos θ  . cos ϕ sin θ cos ψ + sin ϕ sin ψ cos ϕ sin θ sin ψ − sin ϕ cos ψ cos ϕ cos θ Now that’s one hell of a matrix . . . 3 1.2.5 Moving reference frames Let’s examine some point P . This point is described by vector rE in reference frame FE and by rb in reference frame Fb . Also, the origin of Fb (with respect to FE ) is described by the vector rEb . So we have rE = rEb + rb . Now let’s examine the time derivative of rE in FE . We denote this by drE dt = E drEb dt + E drb dt . drE dt E . It is given by (1.2.7) E Let’s examine the terms in this equation. The middle term of the above equation simply indicates the movement of Fb , with respect to FE . The right term is, however, a bit more complicated. It indicates the change of rb with respect to FE . But we usually don’t know this. We only know the change of rb in Fb . So we need to transform this term from FE to Fb . Using a slightly difficult derivation, it can be shown that drb drb = + ΩbE × rb . (1.2.8) dt E dt b The vector ΩbE denotes the rotation vector of Fb with respect to FE . Inserting this relation into the earlier equation gives us drEb drb drE = + + ΩbE × rb . (1.2.9) dt E dt E dt b This is quite an important relation, so remember it well. By the way, it holds for every vector. So instead of the position vector r, we could also take the velocity vector V. Finally, we note some interesting properties of the rotation vector. Given reference frames 1, 2 and 3, we have Ω12 = −Ω21 and Ω31 = Ω32 + Ω21 . (1.2.10) 4 2. Deriving the equations of motion The flight dynamics of an aircraft are described by its equations of motion (EOM). We are going to derive those equations in this chapter. 2.1 2.1.1 Forces The basic force equation To derive the equations of motion of an aircraft, we start by examining forces. Our starting point in this is Newton’s second law. However, Newton’s second law only holds in an inertial reference system. Luckily, the assumptions we have made earlier imply that the Earth-fixed reference frame FE is inertial. (However, Fb is not an inertial reference frame.) So we will derive the equations of motion with respect to FE . Let’s examine an aircraft. Newton’s second law states that Z  Z d Vp dm , F = dF = dt (2.1.1) where we integrate over the entire body. It can be shown that the right part of this equation equals d dt (VG m), where VG is the velocity of the center of gravity of the aircraft. If the aircraft has a constant mass, we can rewrite the above equation into F=m dVG = mAG . dt (2.1.2) This relation looks very familiar. But it does imply something very important. The acceleration of the CG of the aircraft does not depend on how the forces are distributed along the aircraft. It only depends on the magnitude and direction of the forces. 2.1.2 Converting the force equation There is one slight problem. The above equation holds for the FE reference frame. But we usually work in the Fb reference frame. So we need to convert it. To do this, we can use the relation AG = dVG dt = E dVG dt b + ΩbE × VG . (2.1.3) Inserting this into the above equation will give F=m dVG dt b + m ΩbE × VG By the way, in the above equation, we have used that   u   and VG =  v  w   u̇ + qw − rv   = m v̇ + ru − pw . ẇ + pv − qu ΩbE   p   = q  . r (2.1.4) (2.1.5) Here, u, v and w denote the velocity components in X, Y and Z direction, respectively. Similarly, p, q and r denote rotation components about the X, Y and Z axis, respectively. 5 2.1.3 External forces Let’s take a look at the forces F our aircraft is subject to. There are two important kinds of forces: gravity and aerodynamic forces. The gravitational force Fgravity is, in fact, quite simple. It is given by h FE gravity = 0 0 mg iT , (2.1.6) where g is the gravitational acceleration. The superscript E indicates that the force is given in the FE reference frame. However, we want the force in the Fb reference frame. Luckily, we know the transformation matrix TbE . We can thus find that   − sin θ   E (2.1.7) Fb gravity = TbE Fgravity = mg  sin ϕ cos θ  . cos ϕ cos θ The aerodynamic forces Faero are, however, a lot more difficult. For now, we won’t examine them in depth. Instead, we simply say that iT h b b b . (2.1.8) Fb = X Y Z aero By combining this knowledge with the equation of motion for forces, we find that       u̇ + qw − rv − sin θ Xb      b m v̇ + ru − pw = mg  sin ϕ cos θ  +  Y  . Zb ẇ + pv − qu cos ϕ cos θ 2.2 2.2.1 (2.1.9) Moments Angular momentum Before we’re going to look at moments, we will first examine angular momentum. The angular momentum of an aircraft BG (with respect to the CG) is defined as Z BG = dBG = r × VP dm, (2.2.1) where we integrate over every point P in the aircraft. We can substitute V P = VG + dr dt b + ΩbE × r. (2.2.2) If we insert this, and do a lot of working out, we can eventually find that BG = IG ΩbE . The parameter IG is the inertia tensor, with respect to the CG. It is defined as   R  R R Ix −Jxy −Jxz (ry2 + rz2 )dm − (rx ry )dm − (rx rz )dm R 2 R    R  IG = −Jxy Iy −Jyz  =  − (rx ry )dm (rx + rz2 )dm − (ry rz )dm  . R R R 2 −Jxz −Jyz Iz − (rx rz )dm − (ry rz )dm (rx + ry2 )dm (2.2.3) (2.2.4) We have assumed that the XZ-plane of the aircraft is a plane of symmetry. For this reason, Jxy = Jyz = 0. This simplifies the inertia tensor a bit. 6 2.2.2 The moment equation It is now time to look at moments. We again do this from the inertial reference frame FE . The moment acting on our aircraft, with respect to its CG, is given by Z Z Z d (Vp dm) , (2.2.5) MG = dMG = r × dF = r × dt where we integrate over the entire body. Luckily, we can simplify the above relation to MG = dBG dt . (2.2.6) E The above relation only holds for inertial reference frames, such as FE . However, we want to have the above relation in Fb . So we rewrite it to MG = dBG dt b + ΩbE × BG . (2.2.7) By using BG = IG ΩbE , we can continue to rewrite the above equation. We eventually wind up with MG = IG dΩbE dt b + ΩbE × IG ΩbE . In matrix-form, this equation can be written as   Ix ṗ + (Iz − Iy )qr − Jxz (pq + ṙ)   MG = Iy q̇ + (Ix − Iz )pr + Jxz (p2 − r2 ) . Iz ṙ + (Iy − Ix )pq + Jxz (qr − ṗ) (2.2.8) (2.2.9) Note that we have used the fact that Jxy = Jyz = 0. 2.2.3 External moments Let’s take a closer look at MG . Again, we can distinguish two types of moments, acting on our aircraft. There are moments caused by gravity, and moments caused by aerodynamic forces. Luckily, the moments caused by gravity are zero. (The resultant gravitational force acts in the CG.) So we only need to consider the moments caused by aerodynamic forces. We denote those as h iT Mb (2.2.10) N . G,aero = L M This turns the moment equation into     L Ix ṗ + (Iz − Iy )qr − Jxz (pq + ṙ)   2 2 = Iy q̇ + (Ix − Iz )pr + Jxz (p − r ) M  . N Iz ṙ + (Iy − Ix )pq + Jxz (qr − ṗ) 2.3 2.3.1 (2.2.11) Kinematic relations Translational kinematics Now that we have the force and moment equations, we only need to find the kinematic relations for our aircraft. First, we examine translational kinematics. This concerns the velocity of the CG of the aircraft with respect to the ground. 7 The velocity of the CG, with respect to the ground, is called the kinematic velocity Vk . In the FE reference system, it is described by h Vk = VN VE −VZ iT . (2.3.1) In this equation, VN is the velocity component in the Northward direction, VE is the velocity component in the eastward direction, and −VZ is the vertical velocity component. (The minus sign is present because, in the Earth-fixed reference system, VZ is defined to be positive downward.) However, in the Fb reference system, the velocity of the CG, with respect to the ground, is given by h VG = u v w iT . (2.3.2) To relate those two vectors to each other, we need a transformation matrix. This gives us Vk = TEb VG = TTbE VG . (2.3.3) This is the translational kinematic relation. We can use it to derive the change of the aircraft position. To do that, we simply have to integrate the velocities. We thus have Z t Z t Z t −VZ dt. (2.3.4) VE dt and h(t) = VN dt, y(t) = x(t) = 0 2.3.2 0 0 Rotational kinematics Now let’s examine rotational kinematics. This concerns the rotation of the aircraft. In the FE reference system, the rotational velocity is described by the variables ϕ̇, θ̇ and ψ̇. However, in the Fb reference system, the rotational velocity is described by p, q and r. The relation between these two triples can be shown to be      ϕ̇ 1 0 − sin θ p      (2.3.5) q  = 0 cos ϕ cos θ sin ϕ   θ̇  . ψ̇ 0 − sin ϕ cos θ cos ϕ r This is the rotational kinematic relation. q = θ̇ and r = ψ̇. By the way, we can also    1 ϕ̇     θ̇  = 0 0 ψ̇ It is interesting to note that, if ϕ = θ = ψ = 0, then p = ϕ̇, invert the above relation. We would then get   p sin ϕ tan θ cos ϕ tan θ   (2.3.6) cos ϕ − sin ϕ  q  . r sin ϕ/ cos θ cos ϕ/ cos θ 8 3. Rewriting the equations of motion The equations of motion are quite difficult to deal with. To get some useful data out of them, we need to make them a bit simpler. For that, we first linearize them. We then simplify them. And after that, we set them in a non-dimensional form. 3.1 3.1.1 Linearization The idea behind linearization Let’s suppose we have some non-linear function f (X). Here, X is the state of the system. It contains several state variables. To linearize f (X), we should use a multi-dimensional Taylor expansion. We then get f (X) ≈ f (X0 ) + fX1 (X0 )∆X1 + fX2 (X0 )∆X2 + . . . . . . + fXn (X0 )∆Xn + higher order terms. (3.1.1) Here, X0 is the initial point about which we linearize the system. The linearization will only be valid close to this point. Also, the term ∆Xi indicates the deviation of variable Xi from the initial point X0 . When applying linearization, we always neglect higher order terms. This significantly simplifies the equation. (Although it’s still quite big.) 3.1.2 Linearizing the states Now let’s apply linearization to the force and moment equations. We start at the right side: the states. We know from the previous chapter that Fx Fy Fz = m(u̇ + qw − rv), = m(v̇ + ru − pw), = m(ẇ + pv − qu). (3.1.2) (3.1.3) (3.1.4) So we see that Fx = f (u̇, q, w, r, v). The state vector now consists of five states. By applying linearization, we find that Fx Fy Fz = m(u˙0 + q0 w0 − r0 v0 ) + m(∆u̇ + q0 ∆w + w0 ∆q − r0 ∆v − v0 ∆r), = m(v˙0 + r0 u0 − p0 w0 ) + m(∆v̇ + r0 ∆u + u0 ∆r − p0 ∆w − w0 ∆p), = m(w˙0 + p0 v0 − q0 u0 ) + m(∆ẇ + p0 ∆v + v0 ∆p − q0 ∆u − u0 ∆q). (3.1.5) (3.1.6) (3.1.7) We can apply a similar trick for the moments. This would, however, give us quite big expressions. And since we don’t want to spoil too much paper (safe the rainforests!), we will not derive those here. Instead, we will only examine the final result in the end. 3.1.3 Linearizing the forces Now let’s try to linearize the forces. Again, we know from the previous chapter that Fx Fy Fz = −W sin θ + X, = W sin ψ cos θ + Y, = W cos ψ cos θ + Z, 9 (3.1.8) (3.1.9) (3.1.10) where the weight W = mg. We see that this time Fx = f (θ, X). Also, Fy = f (ψ, θ, Y ) and FZ = f (ψ, θ, Z). It may seem that linearization is easy this time. However, there are some problems. The problems are the forces X, Y and Z. They are not part of the state of the aircraft. Instead they also depend on the state of the aircraft. And they don’t only depend on the current state, but on the entire history of states! (For example, a change in angle of attack could create disturbances at the wing. These disturbances will later result in forces acting on the tail of the aircraft.) How do we put this into equations? Well, we say that X is not only a function of the velocity u, but also of all its derivatives u̇, ü, . . .. And the same goes for v, w, p, q and r. This gives us an infinitely big equation. (Great. . . .) But luckily, experience has shown that we can neglect most of these time derivatives, as they aren’t very important. There are only four exceptions. v̇ strongly influences the variables Y and N . Also, ẇ strongly influences Z and M . We therefore say that Fx = f (θ, X) Fy = f (ψ, θ, Y ) Fz = f (ψ, θ, Z) with with with X = f (u, v, w, p, q, r, δa , δe , δr , δt ), Y = f (u, v, w, v̇, p, q, r, δa , δe , δr ), Z = f (u, v, w, ẇ, p, q, r, δa , δe , δr , δt ). (3.1.11) (3.1.12) (3.1.13) When creating the Taylor expansion, we have to apply the chain rule. We then find that Fx (X) ≈ Fx (X0 ) − W cos θ0 ∆θ + Xu ∆u + Xv ∆v + Xw ∆w + Xp ∆p + . . . + Xδt ∆δt , (3.1.14) Fy (X) ≈ Fy (X0 ) − W sin ψ0 sin θ0 ∆θ + W cos ψ0 cos θ0 ∆ψ + Yv̇ ∆v̇ + . . . + Yδr ∆δr , (3.1.15) Fz (X) ≈ Fz (X0 ) − W cos ψ0 sin θ0 ∆θ − W sin ψ0 cos θ0 ∆ψ + Zẇ ∆ẇ + . . . + Zδt ∆δt . (3.1.16) Now that’s one big Taylor expansion. And we haven’t even written down all terms of the equation. (Note the dots in the equation.) By the way, the term Xu indicates the derivative ∂X/∂u. Similarly, Xv = ∂X/∂v, and so on. You may wonder what δa , δe , δr and δt are. Those are the settings of the aileron, elevator, rudder and thrust. These settings of course influence the forces acting on the aircraft. We will examine those coefficients later in more detail. (You may also wonder, why doesn’t Y depend on the thrust setting δt ? This is because we assume that the direction of the thrust vector lies in the plane of symmetry.) 3.2 3.2.1 Simplification Symmetry and asymmetry Let’s try to simplify that monstrocity of an equation of the previous part. To do that, we have to apply several tricks. The most important one, is that of symmetry and asymmetry. We can make a distinction between symmetric and asymmetric forces/deviations. The symmetric deviations (the deviations which don’t break the symmetry) are u, w and q. The symmetric forces/moments are X, Z and M . Similarly, the asymmetric deviations are v, p and r. The asymmetric forces/moments are Y , L and N . It can now be shown that there is no coupling between the symmetric and the asymmetric properties. (That is, as long as the deviations are small.) In other words, X is uneffected by v, p and r. Thus Xv = Xp = Xr = 0. The same trick works for the other forces and moments as well. This causes a lot of terms to disappear in the force equations. 10 3.2.2 Simplifying the force equations There is also another important trick we use, when simplifying the force equations. We assume that the aircraft is flying in a steady symmetric flight. This means that u0 6= 0 u̇0 = 0 v0 = 0 v̇0 = 0 w0 6= 0 ẇ0 = 0 p0 = 0 ṗ0 = 0 ϕ0 = 0 ϕ̇0 = 0 X0 6= 0 q0 = 0 q̇0 = 0 θ0 = 6 0 θ̇0 = 0 Y0 = 0 r0 = 0 ṙ0 = 0 ψ0 = 6 0 ψ̇0 = 0 Z0 6= 0 Ẋ0 = 0, Ẏ0 = 0, Ż0 = 0. (3.2.1) This greatly simplifies the Fx (X0 ), Fy (X0 ) and Fz (X0 ) terms. Now it is finally time to apply all these simplifications and tricks. It will give us the force equations for small deviations from a steady symmetric flight. These equations are −W cos θ0 θ + Xu u + Xw w + Xq q + Xδe δe + Xδt δt W cos θ0 ψ + Yv v + Yv̇ v̇ + Yp p + Yr r + Yδa δa + Yδr δr −W sin θ0 θ + Zu u + Zw w + Zẇ ẇ + Zq q + Zδe δe + Zδt δt = m(u̇ + w0 q), = m(v̇ + u0 r − w0 p), = m(ẇ − u0 q). (3.2.2) (3.2.3) (3.2.4) Of these three equations, the first and the third correspond to symmetric motion. The second equation corresponds to asymmetric motion. You may wonder, where did all the ∆’s go to? Well, to simplify our notation, we omitted them. So in the above equation, all variables indicate the displacement from the initial position X0 . Finally, there is one more small simplification we could do. We haven’t fully defined our reference system yet. (We haven’t specified where the X axis is in the symmetry plane.) Now let’s choose our reference system. The most convenient choice is in this case the stability reference frame FS . By choosing this frame, we have u0 = V and w0 = 0. (V is the velocity.) This eliminates one more term. 3.2.3 The moment equations In a similar way, we can linearize and simplify the moment equations. We won’t go through that tedious process. By now you should more or less know how linearization is done. We’ll just mention the results. They are Lv v + Lp p + Lr r + Lδa δa + Lδr δr Mu u + Mw w + Mẇ ẇ + Mq q + Mδe δe + Mδt δt Nv v + Nv̇ v̇ + Np p + Nr r + Nδa δa + Nδr δr = Ix ṗ − Jxz ṙ, = Iy q̇, = Iz ṙ − Jxz ṗ. (3.2.5) (3.2.6) (3.2.7) Of these three equations, only the second one corresponds to symmetric motion. The other two correspond to asymmetric motion. 3.2.4 The kinematic relations The kinematic relations can also be linearized. (This is, in fact, not that difficult.) After we apply the simplifications, we wind up with ϕ̇ θ̇ ψ̇ = p + r tan θ0 , = q, r . = cos θ0 (3.2.8) (3.2.9) (3.2.10) Of these three equations, only the second one corresponds to symmetric motion. The other two correspond to asymmetric motion. 11 3.3 3.3.1 Setting the equations in a non-dimensional form The dividing term Aerospace engineers often like to work with non-dimensional coefficients. By doing this, they can easily compare aircraft of different size and weight. So, we will also try to make our equations non-dimensional. But how do we do that? We simply divide the equations by a certain value, making them non-dimensional. The question remains, by what do we divide them? Well, we divide the force equations by 21 ρV 2 S, the symmetric moment equation by 21 ρV 2 Sc̄, the asymmetric moment equations by 21 ρV 2 Sb, the symmetric kinematic equation by V /c̄ and the asymmetric kinematic equations by V /b. Here, S is the wing surface area, c̄ is the mean chord length, and b is the wing span. (Note that we use c̄ for symmetric equations, while we use b for asymmetric equations.) 3.3.2 Defining coefficients Dividing our equations by a big term won’t make them look prettier. To make them still readable, we need to define some coefficients. To see how we do that, we consider the term Xu u. We have divided this term by 12 ρV 2 S. We can now rewrite this term to Xu u 1 2 2 ρV S = Xu 1 2 ρV S u = CXu û. V (3.3.1) In this equation, we have defined the non-dimensional velocity û. There is also the coefficient CXu = Xu /( 12 ρV S). This coefficient is called a stability derivative. We can apply the same trick to other terms as well. For example, we can rewrite the term Xw w to Xw w 1 2 2 ρV S = Xw 1 2 ρV S w = CXα α, V (3.3.2) where the angle of attack α is approximated by α = w/V . We can also rewrite the term Xq q to Xq q 1 2S ρV 2 = Xq 1 2 Sc̄ ρV 2 c̄ dθ = CXq Dc θ. V dt (3.3.3) This time we don’t only see a new coefficient. There is also the differential operator Dc . Another differential operator is Db . Dc and Db are defined as c̄ d b d and Db = . (3.3.4) V dt V dt In this way, a lot of coefficients can be defined. We won’t state the definitions of all the coefficients here. (There are simply too many for a summary.) But you probably can guess the meaning of most of them by now. And you simply have to look up the others. Dc = 3.3.3 The equations of motion in matrix form So, we could now write down a new set of equations, with a lot of coefficients. However, we know that these equations are linear. So, we can put them in a matrix form. If we do that, we will find two interesting matrix equations. The equations for the symmetric motion are given by      −CXδe −CXδt " # CXu − 2µc Dc CXα CZ0 CXq û   α   −C  2µc + CZq −CZδt  CZα + (CZα̇ − 2µc )Dc −CX0 CZu      δe Z δe .   =     θ   0  0 0 −Dc 1 0  δt qc̄ 0 Cmq − 2µc KY2 Dc Cmα + Cmα̇ Dc Cmu −Cmδe −Cmδt V (3.3.5) 12 You may note that, instead of using the subscript M , we use the subscript m. This is just a writing I m . convention. You also haven’t seen the variable KY2 yet. It is defined as KY2 = mc̄y2 . Also, µc = ρSc̄ The equations for the asymmetric motion are given by  CYβ + (CYβ̇ − 2µb )Db  0    Clβ CL − 12 Db 0 0 CYp 1 2 Clp − 4µb KX Db Cnp + 4µb KXZ Db    CYr − 4µb −CYδa β     0  ϕ   0   pb  =    −Clδa Clr + 4µb KXZ Db   2V rb 2 −Cnδa Cnr − 4µb KZ Db 2V  −CYδr " # 0   δa .  −Clδr  δr −Cnδr (3.3.6) Again, note that, instead of using the subscripts L and N , we have used l and n. Also, the slip angle β is defined as β = v/V . Cnβ + Cnβ̇ Db 3.3.4 Equations of motion in state-space form We can also put our equation in state-space form, being ẋ = Ax + Bu and y = Cx + Du. (3.3.7) Here, A is the state matrix, B is the input matrix, C is the output matrix and D is the direct matrix. Since the system is time-invariant, all these matrices are constant. Also, x is the state vector, u is the input vector and y is the output vector. The state-space form has several advantages. First of all, the parameters can be solved for at every time t. (The complicated equations for this are known.) Second, computers are very good at performing simulations, once a situation has been described in state-space form. After some interesting matrix manipulation, the state-space form of the symmetric motions can be derived. The result is        xu xα xθ 0 xδe xδt " # û û˙     α̇   z zδ t   δe    u z α z θ z q   α   zδ e . (3.3.8)   +    = V  θ̇   0      δt 0 0 0 0 θ c̄ q̇c̄ qc̄ mδe mδt mu mα mθ mq V V There are quite some strange new coefficients in this equation. The equations, with which these coefficients are calculated, can be looked up. However, we will not mention those here. You may notice that, in the above equation, we only have the state matrix A and the input matrix B. The matrices C and D are not present. That is because they depend on what output you want to get out of your system. So we can’t generally give them here. They are often quite trivial though. Similarly, the state-space form of the asymmetric motions can be found. This time we have    β̇ yβ  ϕ̇   0     ṗb  =   2V   lβ ṙb 2V nβ yϕ 0 0 0 yp 2V b lp np    β yr 0     0  ϕ   0   pb  +    l δa lr   2V nr rb 2V And that concludes this collection of oversized equations. 13 n δa  y δr " # 0   δa .  l δr  δ r n δr (3.3.9) 4. The aerodynamic center In this chapter, we’re going to focus on the aerodynamic center, and its effect on the moment coefficient Cm . 4.1 4.1.1 Force and moment coefficients Aerodynamic forces Let’s investigate a wing. This wing is subject to a pressure distribution. We can sum up this entire pressure distribution. This gives us a resultant aerodynamic force vector CR . Figure 4.1: The forces and moments acting on a wing. Let’s split up the aerodynamic force vector CR . We can do this in multiple ways. We can split the force up into a (dimensionless) normal force coefficient CN and a tangential force coefficient CT . We can also split it up into a lift force coefficient CL and a drag force coefficient CD . Both methods are displayed in figure 4.1. The relation between the four force coefficients is given by CN CT = CL cos α + CD sin α, = CD cos α − CL sin α. (4.1.1) (4.1.2) The coefficients CL , CN , CT and CD all vary with the angle of attack α. So if the angle of attack changes, so do those coefficients. 4.1.2 The aerodynamic moment Next to the aerodynamic forces, we also have an aerodynamic moment coefficient Cm . This moment depends on the reference position. Let’s suppose we know the moment coefficient Cm(x1 ,z1 ) about some point (x1 , z1 ). The moment coefficient Cm(x2 ,z2 ) about some other point (x2 , z2 ) can now be found using Cm(x2 ,z2 ) = Cm(x1 ,z1 ) + CN x2 − x1 z2 − z 1 − CT . c̄ c̄ (4.1.3) Here, c̄ is the mean aerodynamic chord length. (The mean aerodynamic chord (MAC) can be seen as the ‘average’ chord of the entire 3D wing.) Also, x and z denote the position of the reference point in the vehicle reference frame Fr . We define (x0 , z0 ) to be the position of the leading edge of the MAC. 14 4.2 4.2.1 Important points of the wing The center of pressure Let’s put our reference point (x, z) (for calculating Cm ) on the chord. (We thus set z = z0 .) There now is a certain value of x, for which Cm = 0. This point is called the center of pressure (CP). We denote its coordinates by (xd , z0 ). (The CP is the point where the line of action of CR crosses the chord.) Let’s suppose that we know the moment coefficient Cm(x0 ,z0 ) about the leading edge. We can then find xd using xd − x0 . (4.2.1) Cm(xd ,z0 ) = 0 = Cm(x0 ,z0 ) + CN c̄ Let’s define e = xd − x0 . We can then find that Cm(x0 ,z0 ) e . =− c̄ CN 4.2.2 (4.2.2) Lines and metacenters Let’s examine a wing at a certain angle of attack α. This wing is subjected to a resultant force CR . For all points on the line of action of CR , we have Cm = 0. Now let’s examine all points for which dCm /dα = 0. These points also lie on one line. This line is called the neutral line. The point where this line crosses the MAC (and thus z = z0 ) is called the neutral point. The crossing point of the neutral line and the line of action of CR is called the first metacenter M1 . This point has both Cm = 0 and dCm /dα = 0. Let’s take another look at the neutral line. On this line is a point for which d2 Cm /dα2 = 0. This point is called the second metacenter. It is important to remember that all the lines and points discussed above change as α changes. However, the second metacenter changes only very little. We therefore assume that its position is constant for different angles of attack α. 4.2.3 The aerodynamic center Previously, we have defined the second metacenter. However, in aerodynamics, we usually refer to this point as the aerodynamic center (AC). Its coordinates are denoted by (xac , zac ). The corresponding moment coefficient is written as Cmac . We know that we have dCmac /dα = 0 and d2 Cmac /dα2 = 0. We can use this to find xac and zac . To find xac and zac , we have to differentiate equation (4.1.3) with respect to α. Differentiating it once gives dCm(x0 ,z0 ) dCmac dCN xac − x0 dCT zac − z0 =0= + − . (4.2.3) dα dα dα c̄ dα c̄ (Note that we have used the fact that the position of the AC doesn’t vary with α.) Differentiating it twice gives d2 Cm(x0 ,z0 ) d2 CN xac − x0 d2 CT zac − z0 d2 Cmac = 0 = + − . (4.2.4) dα2 dα2 dα2 c̄ dα2 c̄ We now have two equations and two unknowns. We can thus solve for xac and zac . After this, it is easy to find the corresponding moment coefficient Cmac . And since dCmac /dα = 0, we know that this moment coefficient stays the same, even if α varies. We have just described an analytical method to find the AC. There are also graphical methods to find the AC. We won’t go into detail on those methods though. 15 4.2.4 Simplifications We can make a couple of simplifications. Usually, dCT /dα is rather small compared to dCN /dα. We therefore often neglect the effects of the tangential force coefficient CT . If we do this, we find that the AC lies on the MAC (zac = z0 ). In fact, the AC coincides with the neutral point. Finding the position of the AC has now become a lot easier. We know that zac = z0 . We can use this to show that xac satisfies dCm(x0 ,z0 ) xac − x0 =− . (4.2.5) c̄ dCN Once xac has been determined, we can find the moment coefficient about any other point on the wing. Based on our simplifications, we have Cm (x) = Cmac + CN x − xac . c̄ (4.2.6) We can also see another interesting fact from this equation. If CN = 0, the moment coefficient is constant along the wing. And the value of this moment coefficient is equal to Cmac . In other words, the value of Cmac is the value of Cm when CN = 0. (This rule holds for every reference point.) 4.3 4.3.1 Static stability Stability types Let’s suppose that the aircraft is performing a steady flight. The aircraft is then in equilibrium. This means that the moment coefficient about the center of gravity (CG) must be 0. (Cmcg = 0.) Now let’s suppose that the aircraft gets a small deviation from this steady flight. For example, α increases to α + dα. What happens? Due to the change in angle of attack, Cmcg is no longer zero. Instead, it will get a value of dCmcg . We can now distinguish three cases. • The change in moment dCmcg is in the same direction as dα. We thus have dCmcg /dα > 0. In this case, the moment causes α to diverge away from the equilibrium position. The aircraft is therefore unstable. • The change in moment dCmcg is directed oppositely to dα. We now have dCmcg /dα < 0. In this case, the moment causes α to get back to its equilibrium position. The aircraft is thus stable. • The change in moment dCmcg = 0, and thus also dCmcg /dα = 0. In this case, we are in a new equilibrium position. This situation is called neutrally stable or indifferent. 4.3.2 The position of the center of gravity We just saw that, to have a stable aircraft, we should have dCmcg /dα < 0. It turns out that the position of the CG is very important for this. To see why, we differentiate equation (4.2.6) with respect to α. We find that dCmcg dCN xcg − xac = . (4.3.1) dα dα c̄ In general, dCN /dα > 0. So, to have a stable aircraft, we must have xcg − xac < 0. The aerodynamic center should thus be more to the rear of the aircraft than the CG. (This is also why airplanes have a stabilizing horizontal tailplane: It moves the aerodynamic center to the rear.) 16 4.4 4.4.1 Three-dimensional wings Basic and additional lift distributions Previously, we have only examined 2D wings. We will now examine a 3D wing. The wing has a wing span b. Also, at every point y, the 2D airfoil has its own chord c(y) and lift coefficient cl (y, α). It also has a contribution to the lift. By summing up all these lift contributions, we can find the total lift coefficient CL of the wing. This goes according to 1 CL ρV 2 S = 2 2 Z 0 b/2 1 cl (y) ρV 2 c(y) dy 2 ⇒ CL = 2 Z b/2 cl (y) 0 c(y) dy. c̄ (4.4.1) Note that we have used that S = bc̄. We also have used the assumption that the wing is symmetric, by integrating over only one half of the wing. We can split the lift coefficient distribution cl (y, α) up into two parts. First, there is the basic lift distribution clb (y). This is the lift distribution corresponding to the zero-lift angle of attack αCL =0 . (So clb (y) = cl (y, αCL =0 ).) Per definition, we thus have 2 Z b/2 clb (y) 0 c(y) dy = 0. c̄ (4.4.2) Second, there is the additional lift distribution cla (y, α). This lift distribution takes into account changes in α. It is defined as cla (y, α) = cl (y, α)−clb (y). So, if we have α = αCL =0 , then cla (y, αCL =0 ) = 0 for all y. 4.4.2 The aerodynamic center of a 3D wing You may wonder, what is the use of splitting up the lift distribution? Well, it can be shown that the position of the aerodynamic center of the entire wing x̄ac only depends on cla . In fact, we have x̄ac − x̄0 1 2 = c̄ CL Sc̄ Z b/2 0 cla (y, α) c(y) (xac (y) − x̄0 ) dy. (4.4.3) It is important to note the difference between all the x’s. xac is the position of the AC of the 2D airfoil. x̄ac is the position of the AC of the entire 3D wing. Finally, x̄0 is the position of the leading edge of the MAC. By the way, the above equation only holds for reasonable taper and wing twist angles. For very tapered/twisted wings, the above equation loses its accuracy. Now let’s examine the moment coefficient of the entire wing. This moment coefficient only depends on the moment coefficients cmac and the basic lift distribution clb of the individual airfoils. In fact, it can be shown that ! Z b/2 Z b/2 2 Cmac = clb (y) c(y) (xac (y) − x̄0 ) dy . (4.4.4) cmac (y) c(y)2 dy − Sc̄ 0 0 4.4.3 Effects of the 3D wing shape Let’s investigate how the wing shape effects x̄ac and Cmac . There are several properties that we can give to our 3D wing. • A cambered airfoil. Camber causes the value cmac of the individual airfoils to become more negative. So Cmac also becomes more negative. x̄ac doesn’t really change. 17 • A swept wing. When dealing with swept wings, the term (xac (y) − x̄0 ) becomes important. Wings with high sweep angles Λ tend to have a shifting AC at high angles of attack. Whether this improves the stability or not depends on other parameters as well. • A tapered wing. The taper ratio λ = ct /cr (the ratio of the tip chord and the root chord) slightly influences stability. For swept back wings, a low taper ratio tends to have a stabilizing influence. • A slender wing. The aspect ratio A has only little influence on the position of the AC. However, a slender wing (high A) with a large sweep angle Λ will become unstable at large angles of attack. • A twisted wing. Applying a wing twist angle ε causes the basic lift distribution clb to change. This causes Cmac to change as well. In what way Cmac changes, depends on the direction of the wing twist. Predicting the exact behaviour of the wing is, however, rather difficult. A lot of parameters influence the wing behaviour. So don’t be surprised if the above rules don’t always hold. 18 5. Examining an entire aircraft In the previous chapter, we’ve only considered the wing of an aircraft. Now we’re going to add the rest of the aircraft too. How do the various components of the aircraft influence each other? 5.1 5.1.1 Adding a fuselage Changes in moment coefficient Previously, we have only considered a wing. This wing had a moment coefficient Cmw . Now let’s add a fuselage. The combination of wing and fuselage has a moment coefficient Cmwf . The change in moment coefficient ∆Cm is now defined such that Cmwf = Cmw + ∆Cm . (5.1.1) Let’s take a closer look at this change ∆Cm . What does it consist of? We know that a fuselage in a flow usually has a moment coefficient Cmf . However, the wing causes the flow around the fuselage to change. This also causes a moment coefficient induced on the fuselage, denoted by ∆Cmf i . Finally, the fuselage effects the flow around the wing. There is thus also a factor ∆Cmwi . We thus have ∆Cm = Cmf + ∆Cmf i + ∆Cmwi . (5.1.2) In this equation, the coefficients Cmf and ∆Cmf i are usually considered together as Cmf i . 5.1.2 Effects of the fuselage We can use inviscid incompressible flow theory to examine the fuselage. We then find that the moment coefficient of the fuselage, in the induced velocity field, is Z lf π bf (x)2 αf (x) dx. (5.1.3) Cmf i = Cmf + ∆Cmf i = 2Sc̄ 0 Here, bf (x) is the fuselage width and αf (x) is the (effective) fuselage angle of attack. We also integrate over the entire length lf of the fuselage. If there was only a fuselage (and no wing), then the fuselage would have a constant angle of attack. However, the wing causes the angle of attack to vary. In front of the wing, the flow goes up a bit. Behind the wing, there is a downwash. To deal with these complicated effects, we apply linearization. We thus approximate αf (x) as dαf (x) (α − α0 ). (5.1.4) αf (x) = αf0 + dα Here, α0 is the zero normal force angle of attack αCN =0 . αf0 is the corresponding fuselage angle of attack. By using the above equation, we can find a relation for Cmf i . We get Cmf i = 5.1.3 παf0 2Sc̄ Z lf bf (x)2 dx + 0 π(α − α0 ) 2Sc̄ Z 0 lf bf (x)2 dαf (x) dx. dα (5.1.5) The shift of the aerodynamic center Adding the fuselage causes the aerodynamic center to shift. We know that Cmw = Cmacw + CNw x − xacw c̄ and Cmwf = Cmacwf + CNwf 19 x − xacwf . c̄ (5.1.6) Let’s assume that adding the fuselage doesn’t effect the normal force. Thus CNw = CNwf = CN . In this case, we have xacwf − xacw . (5.1.7) ∆Cm = Cmwf − Cmw = ∆Cmac − CN c̄ We can differentiate this equation with respect to α. From the definition of the AC follows that d(∆Cmac )/dα = 0. If we then also use the fact that dCN /dα = CNα , we find that xacwf − xacw ∆xac 1 d(∆Cm ) = =− . c̄ c̄ CNα dα (5.1.8) Part of this shift is caused by the fuselage, while the other part is caused by the new flow on the wing. The shift in angle of attack, due to the fuselage, is   Z lf 1 π dαf (x) 1 dCmf i ∆xac =− dx. (5.1.9) bf (x)2 =− c̄ C dα C 2Sc̄ dα N N 0 α α fi
The shift due to the flow induced on the wing is denoted by ∆xc̄ac wi . We don’t have a clear equation for this part of the shift. However, it is important to remember that this shift is only significant for swept wings. If there is a positive sweep angle, then the AC moves backward. 5.2 5.2.1 Adding the rest of the aircraft The three parts It is now time to examine an entire aircraft. The CG of this aircraft is positioned at (xcg , zcg ). We split this aircraft up into three parts. • First, there is the wing, with attached fuselage and nacelles. The position of the AC of this part is (xw , zw ). Two forces and one moment are acting in this AC. There are a normal force Nw (directed upward), a tangential force Tw (directed to the rear) and a moment Macw . • Second, we have a horizontal tailplane. The AC of this part is at (xh , zh ). In it are acting a normal force Nh , a tangential force Th and a moment Mach . • Third, there is the propulsion unit. Contrary to the other two parts, this part has no moment. It does have a normal force Np and a tangential force Tp . However, these forces are all tilted upward by the thrust inclination ip . (So they have different directions then the force Tw , Th , Nw and Nh .) Also, the tangential force Tp is defined to be positive when directed forward. (This is contrary to the forces Th and Tw , which are positive when directed backward.) 5.2.2 The equations of motion We will now derive the equations of motion for this simplified aircraft. We assume that the aircraft is in a fully symmetric flight. We then only need to consider three equations of motion. Taking the sum of forces in X direction gives T = Tw + Th − Tp cos ip + Np sin ip = −W sin θ. (5.2.1) Similarly, we can take the sum of forces in Z direction, and the sum of moments about the CG. (This is done about the Y axis.) We then get N = Nw + Nh + Np cos ip + Tp sin ip = W cos θ, 20 (5.2.2) M =Macw + Nw (xcg − xw ) − Tw (zcg − zw ) + Mach + Nh (xcg − xh ) − Th (zcg − zh ) + . . . . . . + (Np cos ip + Tp sin ip )(xcg − xp ) + (Tp cos ip − Np sin ip )(zcg − zp ) = 0. (5.2.3) We can simplify these equations, by making a couple of assumptions. We want to examine the stability of the aircraft. The propulsion doesn’t influence the stability of the aircraft much. So we neglect propulsion effects. We also neglect Th , since it is very small compared to Tw . We assume that (zcg − zw ) ≈ 0. And finally, we neglect Mach . This gives us T = Tw = −W sin θ, (5.2.4) N = Nw + Nh = W cos θ, (5.2.5) M = Macw + Nw (xcg − xw ) + Nh (xcg − xh ) = 0. (5.2.6) That simplifies matters greatly. 5.2.3 Non-dimensionalizing the equations of motion Let’s non-dimensionalize the equations of motion of the previous paragraph. For that, we divide the force equations by 21 ρV 2 S and the moment equation by 21 ρV 2 Sc̄. This then gives us W sin θ, 2S ρV 2 (5.2.7) CT = CTw = − 1 CN = CNw + CNh Cm = Cmacw + CNw  Vh V 2 Sh = S xcg − xw − CNh c̄ W 1 2 2 ρV S  Vh V 2 cos θ, (5.2.8) Sh l h = 0. Sc̄ (5.2.9) A lot of new coefficients have suddenly disappeared. These coefficients are defined, such that 1 N = CN ρV 2 S, 2 1 T = CT ρV 2 S, 2 1 M = Cm ρV 2 Sc̄, 2 (5.2.10) 1 1 1 Tw = CTw ρV 2 S, Macw = Cmacw ρV 2 Sc̄, (5.2.11) Nw = CNw ρV 2 S, 2 2 2 1 1 1 Th = CTh ρVh2 Sh , Mach = Cmach ρVh2 Sh c̄h . (5.2.12) Nh = CNh ρVh2 Sh , 2 2 2 Here, 12 ρVh2 is the average local dynamic pressure on the horizontal tail plain. Also, Sh is the tailplane h lh is known as the tailplane volume. surface area and c̄h is the MAC of the tailplane. The quantity SSc̄ And finally, we have defined the tail length lh = xh − xw ≈ xh − xcg . 5.3 5.3.1 The horizontal tailplane Important angles We will now take a closer look at the horizontal tailplane. There are three parameters that describe the configuration of the horizontal tailplane. These parameters are the effective horizontal tailplane angle of attack αh , the elevator deflection δe and the elevator trim tab deflection δte . The three angles are visualized in figure 5.1. There is one angle which we will examine more closely now. And that is the effective angle of attack αh . It is different from the angle of attack of the aircraft α. There are two important causes for this. First, 21 Figure 5.1: The angles of the horizontal tailplane. the horizontal tail plane has an incidence angle ih , relative to the MAC of the wing. And second, the tailplane experiences downwash, caused by the wing of the aircraft. The average downwash angle is denoted by ε. By putting this all together, we find that αh = α + ih − ε. (5.3.1) We can elaborate a bit further on this. The downwash ε mainly depends on α. Linearization thus gives dε (α − α0 ). It follows that ε ≈ dα   dε αh = 1 − (α − α0 ) + (α0 + ih ). (5.3.2) dα From this follows that the derivate dαh /dα is given by dε dαh =1− . dα dα (5.3.3) This derivative is thus generally smaller than 1. 5.3.2 The horizontal tailplane normal force Let’s examine the normal force CNh of the horizontal tailplane. This is a function of the three angles αh , δe and δte . Applying linearization gives CNh = CNh0 + ∂CNh ∂CNh ∂CNh αh + δe + δe . ∂αh ∂δe ∂δet t (5.3.4) The effect of the trim tab to the normal force is usually negligible. So, ∂CNh /∂δet ≈ 0. Also, since most horizontal tailplanes are (nearly) symmetric, we have CNh0 ≈ 0. This simplifies the above equation to CNh = ∂CNh ∂CNh αh + δe = CNhα αh + CNhδ δe . ∂αh ∂δe (5.3.5) Note that we have used a shorter notation in the right part of the above equation. The variables CNhα and CNhδ are quite important for the balance of the control surface. If they are both negative, then the control surface is called aerodynamically underbalanced. If, however, they are both positive, then the control surface is called aerodynamically overbalanced. 5.3.3 The elevator deflection necessary for equilibrium We can ask ourselves, what elevator deflection δe should we have, to make sure our aircraft is in equilibrium? For that, we examine the moment equation (5.2.9). In this equation are the coefficients CNw and 22 CNh . We can replace these by the linearizations CNw = CNwα (α − α0 ) and CNh = CNhα αh + CNhδ δe . (5.3.6) If we do this, we find that Cm = Cmacw  xcg − xw  − CNhα αh + CNhδ δe + CNwα (α − α0 ) c̄  Vh V 2 Sh l h = 0. Sc̄ (5.3.7) We can now also substitute the relation (5.3.2) for αh . Doing this, and working the whole equation out, gives (5.3.8) Cm = Cm0 + Cmα (α − α0 ) + Cmδe δe = 0, where Cmα is known as the static longitudinal stability and Cmδe is the elevator effectivity. Together with the constant Cm0 , they are defined as Cm0 Cmα Cmδe xcg − xw − CNhα c̄  2 Sh l h Vh = −CNhδ . V Sc̄ = CNwα  2 Sh l h , Sc̄   2  Vh Sh l h dε , 1− dα V Sc̄ = Cmacw − CNhα (α0 + ih ) Vh V (5.3.9) (5.3.10) (5.3.11) We can now solve for δe . It is simply given by δe = − Cm0 + Cmα (α − α0 ) . Cmδe (5.3.12) This is a nice expression. But do remember that we have made several linearizations to derive this equation. The above equation is thus only valid, when all the linearizations are allowed. 23 6. Longitudinal stability derivatives We have seen a lot of stability derivatives in previous chapters. However, it would be nice to know their values. We’re therefore going to derive some relations for them. In this chapter, we will look at longitudinal stability derivatives. In the next chapter, we’ll examine lateral stability derivatives. 6.1 6.1.1 Nominal stability derivatives Methods of finding the stability derivatives There are three methods to find the stability derivatives. The first one is of course by performing flight tests or wind tunnel tests. These tests are, however, quite expensive. An alternative is using computational fluid dynamics (CDF). This is usually less expensive, but it still requires a lot of work. Finally, simple analytic expressions can be used. They are usually not very accurate. (Especially not for strange aircraft configurations.) However, they don’t require a lot of work. In this chapter, we’re going to examine these analytic expressions. 6.1.2 Equations of motion Before we will find stability derivatives, we first need to re-examine the equations of motion. The symmetric equations of motion, for an airplane in a steady flight, are X = − D cos α + L sin α + Tp cos(α0 + ip ) = W sin γ0 , Z = − L cos α − D sin α − Tp sin(α0 + ip ) = −W cos γ0 . (6.1.1) (6.1.2) Here, α0 is the initial angle of attack. (It is now not the zero-lift angle of attack.) Also, α now denotes the deviation from this flight condition. We assume α0 + ip is small, so we can use the small angle approximation. If we also non-dimensionalize the above relations, we find that CX = − CD cos α + CL sin α + Tc′ = W 1 2 2 ρV S sin γ0 , (6.1.3) W cos γ0 , 2 2 ρV S CZ = − CL cos α − CD sin α − Tc′ (α0 + ip ) = − 1 where we have defined Tc′ = 6.1.3 (6.1.4) Tp . 1 2 2 ρV S (6.1.5) Nominal flight conditions Let’s examine an aircraft flying a steady horizontal flight. We will now try to find the nominal stability derivatives CX0 , CZ0 and Cm0 . Since the aircraft is flying horizontally, we have α = γ0 = 0. (Remember that α is the deviation from the steady flight.) The relations of the previous paragraph now turn into CX0 = −CD + Tc′ = 0 and CZ0 = −CL − Tc′ (α0 + ip ) = Finally, from moment equilibrium follows that Cm0 = 0. 24 W . 1 2 2 ρV S (6.1.6) 6.2 6.2.1 Velocity stability derivatives The basic relations Now let’s find the stability derivatives with respect to the velocity. They are CXu , CZu and Cmu . They are very hard to determine experimentally. This is because wind tunnels and flying aircraft can’t change their velocity in a very accurate way. Luckily, we can find expressions for them. Let’s start to examine CXu . We can recall that CXu = 1 1 2 ρV ∂X S ∂V 1 X = CX ρV 2 S. 2 and (6.2.1) (We have used the fact that ∂V /∂u ≈ 1.) Taking the derivative of the second relation, with respect to V , gives ∂CX 1 2 ∂X = CX ρV S + ρV S. (6.2.2) ∂V ∂V 2 Inserting this into the first relation will give CXu = 2CX + ∂CX V. ∂V (6.2.3) In a similar way, we can find the expressions for CZu and Cmu . They are CZu = 2CZ + 6.2.2 ∂CZ V ∂V and Cmu = 2Cm + ∂Cm V. ∂V (6.2.4) Rewriting the relations There are still some terms we don’t know in the relations of the previous paragraph. They are CX , CZ , Cm , ∂CX /∂V , ∂CZ /∂V and ∂Cm /∂V . How can we rewrite them? We are considering deviations from the steady horizontal flight. So we can replace CX by CX0 = −CD +Tc′ . Similarly, CZ is replaced by CZ0 = −CL − Tc′ (α0 + ip ) and Cm by Cm0 = 0. That simplifies the equations quite a bit. The derivatives are a bit harder to rewrite. At a steady horizontal flight, we have CX = −CD + Tc′ and CZ = −CL − Tc′ (α0 + ip ). Differentiating this gives ∂CD ∂Tc′ ∂CX =− + ∂V ∂V ∂V ∂CZ ∂CL ∂Tc′ =− − (α0 + ip ). ∂V ∂V ∂V and (6.2.5) That leaves us with some more derivatives. First, let’s examine ∂CD /∂V , ∂CL /∂V and ∂Cm /∂V . To find them, we have to know why the coefficients CD , CL and Cm vary with airspeed. These variations are partly caused by changes in Mach number M , changes in thrust Tc′ and changes in Reynolds number Re. Although changes in the Reynolds number may be neglected, we do have to consider M and Tc′ . This implies that ∂CD ∂Tc′ ∂CD ∂CD ∂M = + , ∂V ∂M ∂V ∂Tc′ ∂V ∂CL ∂CL ∂M ∂CL ∂Tc′ = + , ∂V ∂M ∂V ∂Tc′ ∂V ∂Cm ∂Cm ∂M ∂Cm ∂Tc′ = + . ∂V ∂M ∂V ∂Tc′ ∂V 25 (6.2.6) (6.2.7) (6.2.8) If we use this, in combination with earlier equations, we will find that   ∂CD dTc′ ∂CD ′ CXu = − 2CD + 2Tc + 1 − V − M, ′ ∂Tc dV ∂M   ∂CL ∂CL dTc′ ′ V − M, CZu = − 2CL − 2Tc (α0 + ip ) − (α0 + ip ) + ′ ∂Tc dV ∂M ∂Cm dTc′ ∂Cm Cmu = V + M. ′ ∂Tc dV ∂M 6.2.3 (6.2.9) (6.2.10) (6.2.11) The thrust derivative The equations of the previous paragraph still have a lot of derivatives. We won’t go into detail on derivatives with respect Tc′ or M . However, we will consider dTc′ /dV . It turns out that we can write this derivative as T′ dTc′ = −k c , (6.2.12) dV V where the constant k depends on the flight type. If we have a gliding flight, then Tc′ = 0. Thus also dTc′ /dV = 0 and therefore k = 0. If we have a jet-powered aircraft, then Tp = Tc′ 21 ρV 2 S = constant. From this follows that k = 2. Finally, for propeller-powered aircraft, we have Tp V = Tc′ 21 ρV 3 S = constant. This implies that k = 3. Let’s use the above relation to simplify our equations. The equations for CXu , CZu and Cmu now become    ∂CD ∂CD CXu = − 2CD + 2 − k 1 − Tc′ − M, (6.2.13) ∂Tc′ ∂M   ∂CL ∂CL M, (6.2.14) Tc′ − CZu = − 2CL + (k − 2)(α0 + ip ) + k ∂Tc′ ∂M ∂Cm ′ ∂Cm Cmu = − k T + M. (6.2.15) ∂Tc′ c ∂M When specific data about the type of flight is known, the above equations can be simplified even further. For example, when the flight is at low subsonic velocities, then Mach effects may be neglected. Thus ∂CD /∂M = ∂CL /∂M = ∂Cm /∂M = 0. In other cases, there are often other simplifications that can be performed. 6.3 6.3.1 Angle of attack stability derivatives The basic relations for CXα and CZα We will now try to find relations for CXα , CZα and Cmα . First we examine CXα and CZα . They are defined as ∂CX ∂CZ 1 ∂Z 1 ∂X = and CZα = 1 = . (6.3.1) CXα = 1 ∂w ∂α ∂w ∂α 2 ρV S 2 ρV S If we take the derivative of equations (6.1.3) and (6.1.4), we find that CXα = − CDα cos α + CD sin α + CLα sin α + CL cos α, CZα = − CLα cos α + CL sin α − CDα sin α − CD cos α. (6.3.2) (6.3.3) We are examining an aircraft performing a steady horizontal flight. Thus α = 0. This simplifies the above equations to CXα = CL − CDα and CZα = −CLα − CD ≈ −CLα ≈ −CNα . In the last part of the above equation, we have used the fact that CD is much smaller than CLα . 26 (6.3.4) 6.3.2 Rewriting the relation for CXα We can try to rewrite the relation for CXα . To do this, we examine CDα . Let’s assume that the aircraft has a parabolic drag curve. This implies that CD = CD0 + CL2 , πAe which, in turn, implies that CDα = 2 CLα CL . πAe If we combine this with the former expression for CXα , we wind up with   CL CXα = CL 1 − 2 α . πAe 6.3.3 (6.3.5) (6.3.6) The relation for Cmα In a previous chapter, we have already considered Cmα . After neglecting the effects of many parts of the aircraft, we wound up with   2  xcg − xw Vh Sh l h dε Cmα = CNwα − CNhα 1 − . (6.3.7) c̄ dα V Sc̄ 6.4 6.4.1 Pitch rate stability derivatives The reasons behind the changing coefficients We will now try to find CXq , CZq and Cmq . Luckily, CX doesn’t get influenced a lot by q. So it is usually assumed that CXq = 0. That saves us some work. We now only need to find CZq and Cmq . They are defined as ∂M 1 ∂CZ ∂Cm 1 ∂Z CZq = 1 = qc̄ = qc̄ . and Cmq = 1 (6.4.1) 2 ∂q ∂q ∂V ∂V 2 ρV Sc̄ 2 ρV Sc̄ To find CZq and Cmq , we first have to understand some theory behind rotations. Why do the coefficients change, when the aircraft rotates? This is because the effective angle of attack changes. Imagine an aircraft with its nose pitching upward. The tailplane of the aircraft is thus pitching downward. Now imagine you’re sitting on the tailplane. As seen from the tailplane, it looks like the flow of air is coming upward. This means that the tailplane experiences a bigger angle of attack. To find the exact value of the change in angle of attack ∆α, we examine the center of rotation. This is the point about which the aircraft appears to be rotating. The center of rotation lies on the Zs -axis. The apparent rotation itself is performed with a radius R, which is given by R= V . q (6.4.2) The change in angle of attack ∆α, at any point x on the airplane, is then given by ∆α = 6.4.2 x − xcg qc̄ x − xcg = . R c̄ V (6.4.3) The changing coefficients We know that the apparent angle of attack changes across the aircraft. This is especially important for the horizontal tailplane. In fact, the change in angle of attack of the tailplane is given by ∆αh = xh − xcg qc̄ lh qc̄ ≈ . c̄ V c̄ V 27 (6.4.4) This change in angle of attack causes the normal force of the tailplane to change. In fact, it changes by an amount  2  2 Sh Sh lh qc̄ Vh Vh ∆αh = CNhα . (6.4.5) ∆CNh = CNhα V S V Sc̄ V Similarly, the change of the moment is given by  2  2 Sh l h Sh lh2 qc̄ Vh Vh ∆αh = −CNhα . ∆Cm = −CNhα V Sc̄ V Sc̄2 V (6.4.6) qc̄ We know that CZq = ∂CZ /∂ qc̄ V and Cmq = ∂Cm /∂ V . By using this, we can find the contributions of the horizontal tailplane to Czq and Cmq . They are  2  2

Sh l h Sh lh2 Vh Vh and Cmq h = −CNhα . (6.4.7) CZq h = −CNhα V Sc̄ V Sc̄2 (The minus sign in the left part appeared, because CN is defined upward, while CZ is defined downward.) There is, however, one small problem. The aircraft doesn’t consist of only a horizontal tailplane. It also has various other parts. But it is very difficult to calculate the effects of all these parts. For that reason, we make an estimate. We say
that the contribution of the full aircraft CZq is twice the contribution of the horizontal tailplane CZq h . This implies that CZq = 2 CZq
h = 2CNhα  Vh V 2 Sh l h . Sc̄ (6.4.8) For Cmq , we apply the same trick. But instead of a factor 2, a factor between 1.1 to 1.2 should now be used, depending on the aircraft. We thus get  2
Sh lh2 Vh Cmq = (1.1 ∼ 1.2) Cmq h = −(1.1 ∼ 1.2)CNhα . (6.4.9) V Sc̄2 6.5 6.5.1 Other longitudinal stability derivatives Vertical acceleration stability derivatives We now examine CZα̇ and Cmα̇ . (We assume CXα̇ = 0.) To do this, we look at the horizontal tailplane. During a steady flight, it has an effective angle of attack αh = α − ε + ih = α − dε α + ih . dα (6.5.1) Now let’s suppose that the aircraft experiences a change in angle of attack. This causes the downwash angle ε of the wing to change. A time ∆t = lh /V later will this change be experienced by the horizontal tailplane. In other words, the downwash ε(t) at time t depends on the angle of attack α(t − ∆t) at time t − ∆t. A linear approximation of α(t − ∆t) is given by α(t − ∆t) = α(t) − α̇∆t. (6.5.2) By using this, we find that the downwash is given by ε(t) = dε dε lh dε α(t − ∆t) = α(t) − α̇ . dα dα dα V (6.5.3) This implies that the effective angle of attack is given by αh = α − dε dε lh α+ α̇ + ih . dα dα V 28 (6.5.4) The change in effective angle of attack is ∆αh = dε lh α̇c̄ . dα c̄ V (6.5.5) We now have enough data to find the coefficients CZα̇ and Cmα̇ . We know that ∆CZ = −CNhα ∆Cm = −CNhα   Vh V Vh V 2 2  Sh ∆αh = −CNhα S Vh V 2 Sh lh dε α̇c̄ , Sc̄ dα V  Vh V 2 Vh V 2 Sh lh dε , Sc̄ dα Sh l h ∆αh = −CNhα Sc̄ Sh lh2 dε α̇c̄ . Sc̄2 dα V (6.5.6) (6.5.7) The coefficients CZα̇ and Cmα̇ are now given by CZα̇ = Cmα̇ 6.5.2 1 ∂Z ∂CZ = α̇c̄ = −CNhα 1 ∂ ẇ ∂V 2 ρSc̄  ∂Cm 1 ∂M = α̇c̄ = −CNhα = 1 2 ∂ ẇ ∂V 2 ρSc̄  Vh V 2 Sh lh2 dε . Sc̄2 dα (6.5.8) (6.5.9) Elevator angle stability derivatives The last stability derivatives we will consider in this chapter are CXδe , CZδe and Cmδe . Usually CX doesn’t vary a lot with δe , so we assume that CXδe = 0. But what about CZδe ? Well, this one is given by CZδe = −CNhδ e  Vh V 2 Sh . S (6.5.10) Finally there is Cmδe . We can find that it is Cmδe = CZδe xh − xcg ≈ −CNδe c̄  Vh V 2 Sh l h . Sc̄ (6.5.11) The coefficient CZδe usually isn’t very important. However, Cmδe is very important. This is because the whole goal of an elevator is to apply a moment to the aircraft. 6.5.3 Effects of moving the center of gravity We have one topic left to discuss. What happens when the CG moves from position 1 to position 2? In this case, several coefficients change. This goes according to xcg2 − xcg1 , c̄ xcg2 − xcg1 = CZq1 − CZα , c̄ 2 
xcg2 − xcg1 xcg2 − xcg1 , + CZα = Cmq1 − CZq1 + Cmα1 c̄ c̄ xcg2 − xcg1 . = Cmα̇1 − CZα̇ c̄ Cmα2 = Cmα1 − CZα (6.5.12) CZq2 (6.5.13) Cmq2 Cmα̇2 29 (6.5.14) (6.5.15) 7. Lateral stability derivatives In the previous chapter, we found relations for the longitudinal stability derivatives. Now we’ll examine the lateral stability derivatives. 7.1 7.1.1 Sideslip angle stability derivatives Horizontal forces We start by examining derivatives with respect to the sideslip angle β. This angle is defined as v v ≈ . (7.1.1) β = arcsin V V We will now examine CYβ . Let’s examine an aircraft with a sideslip angle β. This sideslip angle causes a horizontal force Y on the aircraft. The most important contributors to this horizontal force are the fuselage and the vertical tailplane. First let’s examine the vertical tailplane. Luckily, this tailplane has a lot of analogies with the horizontal tailplane, so we can use some short cuts. For example, the force acting on the vertical tailplane is given by  2
dαv Vv Sv CYβ v = CYvα , (7.1.2) dβ V S where Sv is the vertical tailplane surface area and Vv is the average velocity over it. Also, CYvα = ∂CYv ∂αv . Let’s take a closer look at the effective angle of attack of the tailplane αv . It’s not equal to β. This is because the fuselage also alters the flow by an angle σ. (σ is similar to the downwash ε for the horizontal tailplane.) The vertical tailplane thus has an angle of attack of αv = −(β − σ). (The minus is present due to sign convention.) Inserting this relation into the above equation gives   2 
Vh Sv dσ . (7.1.3) CYβ v = −CYvα 1 − dβ V S Usually, most terms in the above equation are known. Only dσ/dβ is still a bit of a mystery. It is very hard to determine. However, it usually is negative. (So dσ/dβ < 0.) Next to the tailplane contribution, there is usually also a contribution by the fuselage. However, we don’t go into depth on that here. 7.1.2 Rolling momemts Now let’s examine the so-called effective dihedral Clβ . The coefficient Cl was defined as Cl = L . 1 2 2 ρV b (7.1.4) It is important to note that L is not the lift. It is the moment about the X axis. Cl is thus not the lift coefficient either. The effective dihedral Clβ mostly depends on the wing set-up. Both the wing-dihedral Γ and the sweep angle Λ strongly effect Cl . (The wing-dihedral Γ is the angle with which the wings have been tilted upward, when seen from the fuselage.) First let’s examine an aircraft with a wing-dihedral Γ. We suppose that the aircraft is sideslipping to the right. From the aircraft, it now appears as if part of the flow is coming from the right. This flow 30 ‘crouches’ under the right wing, pushing it more upward. However, it flows over the left wing, pushing that one downward a bit. This thus causes the aircraft to roll to the left. To find more info about the moment caused by the wing-dihedral, we need to examine the new angle of attacks of the wings αwl and αwr . By using small angle approximations, we can find that αwl ≈ α − βΓ and αwr ≈ α + βΓ. (7.1.5) The changes in the angles of attack are thus ∆αwl = −βΓ and ∆αwr = βΓ. So the moment caused by the wing-dihedral is approximately linearly dependend on both β and Γ. (We thus have Clβ ∼ Γ.) Second, we look at an aircraft with a wing sweep angle Λ. The lift of a wing strongly depends on the flow velocity perpendicular to the leading edge. Again, we suppose that part of the flow is coming in from the right. This causes the flow to be more perpendicular w.r.t. to the right wing leading edge, thus increasing the lift. However, the flow is more parallel w.r.t. the leading edge of the left wing. The left wing thus has reduced lift. It can be shown that the change in lift for the aircraft, due to a sweep angle Λ, is
1 S 1 cos2 (Λ − β) − cos2 (Λ + β) ≈ CL ρV 2 S sin (2Λβ) . (7.1.6) ∆L = CL ρV 2 2 2 2 The rightmost part of the equation is an approximation. It only works for small values of β. The above equation shows that the lift more or less linearly depends on Λ and β. It can be shown that the same holds for the moment Cl . The effective dihedral Clβ is thus proportional to Λ. Next to the wing, also the horizontal tailplane and the fuselage effect Clβ . However, we won’t examine these effects. 7.1.3 Yawing moments The stability derivative Cnβ is called the static directional stability. (It’s also known as the Weathercock stability.) It is about just as important as Cmα . It can be shown that, if Cnβ is positive, then the aircraft is stable for yawing motions. However, if Cnβ is negative, then the aircraft is unstable for yawing motions. Naturally, we want to have Cnβ > 0. Luckily, the wings and the horizontal tailplane have a slightly positive effect on Cnβ . However, the fuselage causes Cnβ to decrease. To compensate for this, a vertical tailplane is used, strongly increasing Cnβ . Let’s examine the effects of this tailplane. You may remember that the normal force on it was    2
dσ Vv Sv CYβ v = −CYvα 1 − . dβ V S This normal force causes a moment  
zv − zcg
xv − xcg Cnβ v = − CYβ v sin α0 + cos α0 . b b (7.1.7) (7.1.8) z −z We can usually assume α0 to be small. (Thus cos α0 ≈ 1.) Also, v b cg sin α0 is usually quite small, compared to the other term, so we neglect it. If we also use the tail length of the vertical tailplane lv = xv − xcg , we can rewrite the above equation to   2 
Vv Sv l v dσ . (7.1.9) Cnβ v = CYvα 1 − dβ V Sb From this, the correspondence to Cmα again becomes clear. To emphasize this, we
once more show the equation for the horizontal tailplane contribution to Cmα . Rather similar to Cnβ v , it was given by    2 dε Vh Sh l h (Cmα )h = −CNhα 1 − . (7.1.10) dα V Sc̄ 31 7.2 7.2.1 Roll rate stability derivatives Horizontal forces It is time to investigate the effects of roll. In other words, we will try to find the stability derivatives CYp , Clp and Cnp . (Of these three, Clp is the most important.) First we examine CYp . It is defined such that Yp = CYp pb 1 2 ρV S. 2V 2 (7.2.1) The only part having a more or less significant contribution to CYp is the vertical tailplane. Let’s examine a rolling aircraft. Due to this rolling, the vertical tailplane is moving horizontally. It will therefore get an effective angle of attack. This causes a horizontal force. A positive roll rate gives a negative horizontal force. CYp is thus negative. However, CYp is usually rather small. For this reason it is often neglected. So we say that CYp ≈ 0. 7.2.2 Rolling moments Now we will try to find Clp . Again, we examine a rolling aircraft. One wing of the aircraft goes up, while the other one goes down. This motion changes the effective angle of attack and thus also the lift of the wings. The upward going wing will get a lower lift, while the downward moving wing will experience a bigger amount of lift. The wing forces thus cause a moment opposite to the rolling motion. This means that Clp is highly negative. It also implies that the rolling motion is very strongly damped. (We will see this again, when examining the aperiodic roll in chapter 10.) We can also investigate the actual effects of the rolling motion. To do this, we examine a chord at a distance y from the fuselage. This chord will have an additional vertical velocity of py. The change in angle of attack of this chord thus is py pb y ∆α = = . (7.2.2) V 2V b/2 So a chord that is far away from the fuselage will experience a big change in angle of attack. The change in lift is therefore biggest for these chords. These chords also have a relatively big distance to the CG of the aircraft. For this reason, they will significantly effect the resulting moment. Other parts of the aircraft may also influence Clp slightly. However, their influence is very small, compared to the effects of the wings. The contributions of the other parts are therefore neglected. 7.2.3 Yawing moments To find Cnp , we again examine a rolling aircraft. The rolling of the aircraft has two important effects. First, we look at the vertical tailplane. As was discussed earlier, this tailplane will move. It thus has an effective angle of attack, and therefore a horizontal force. This horizontal force causes the aircraft to yaw. A positive rolling motion causes
a positive yawing moment. The vertical tailplane thus has a positive contribution to Cnp . (So Cnp v > 0.) But now let’s look at the wings. Let’s suppose that the aircraft is rolling to the right. For the right wing, it then appears as if the flow comes (partially) from below. The lift is per definition perpendicular to the direction of the incoming flow. The lift vector is thus tilted forward. Part of this lift causes the aircraft to yaw to the left. The opposite happens for the left wing: The lift vector is tilted backward. Again, this causes a yawing moment to the left. (This effect is known as adverse yaw.) So we conclude that,
due to the wings, a positive rolling motion results in a negative yawing moment. We thus have Cnp w < 0. 32 For most normal flights, the effects of the vertical stabilizer are a bit bigger than the effects of the wing. We thus have Cnp > 0. However, high roll rates and/or high angles of attack increase the effect of the wings. In this case, we will most likely have Cnp < 0. 7.3 7.3.1 Yaw rate stability derivatives Horizontal forces In this part, we’ll try to find the stability derivatives CYr , Clr and Cnr . We start with the not very important coefficient CYr . Let’s examine a yawing aircraft. Due to the yawing moment, the vertical tailplane moves horizontally. Because of this, its effective angle of attack will change by ∆αv = rb lv rlv = . V 2V b/2 The contribution of the tailplane to CYr is now given by  2 Sv l v Vv (CYr )v = 2CYvα . V Sb (7.3.1) (7.3.2) The contribution is positive, so (CYr )v > 0. Next to the vertical tailplane, there are also other parts influencing CYr . Most parts have a negative contribution to CYr . However, none of these contributions are as big as (CYr )v . The stability derivative CYr is therefore still positive. It is only slightly smaller than (CYr )v . 7.3.2 Rolling moments We will now examine Clr . There are two important contributions to Clr . They come from the vertical tailplane and the wings. First we examine the vertical tailplane. We just saw that a yawing motion causes a horizontal force on the vertical tailplane. This horizontal force causes a moment   xv − xcg zv − zcg (7.3.3) (Clr )v = (CYr )v cos α0 − sin α0 . b b A positive yawing motion results in a positive moment. We thus have (Clr )v > 0. Now let’s examine the wings. Because of the yawing motion, one wing will move faster, while the other wing will move slower. This causes the lift on one wing to increase, while it will decrease on the other wing. This results in a rolling moment. Sadly, it’s rather hard to find an equation for the moment caused by the wings. So we won’t examine that any further. However, it is important to remember that a positive yawing motion causes a positive rolling moment. We thus have (Clr )w > 0. The total coefficient Clr is then, of course, also positive. 7.3.3 Yawing moments Finally, we examine Cnr . The most important contribution comes from the vertical tailplane. We know that a yawing motion causes a horizontal force on the vertical tailplane. This force is such that it damps the yawing motion. The contribution (Cnr )v is thus very highly negative. In fact, it is given by  2 Vv lv Sv lv2 = −2CYvα . (7.3.4) (Cnr )v = − (CYr )v b V Sb2 33 The vertical tailplane is about the only part seriously effecting the coefficient Cnr . Sometimes also the fuselage effects it. This effect is also negative. (So (Cnr )f < 0.) The coefficient Cnr itself is thus also very strongly negative. This implies that the yawing motion is highly damped. 7.4 7.4.1 Other lateral stability derivatives Aileron deflections Let’s consider the ailerons. The aileron deflection δa is defined as δa = δaright − δalef t . (7.4.1) A deflection of the ailerons causes almost no change in horizontal forces. We thus have CYδa = 0. The so-called aileron effectiveness Clδa is, of course, not negligible. (Causing moments about the X axis is what ailerons are for.) The coefficient Cnδa usually isn’t negligible either. By the way, the moments caused by an aileron deflection are given by 1 L = Clδa δa ρV 2 Sb 2 1 N = Cnδa δa ρV 2 Sb. 2 and (7.4.2) Clδa is negative. A positive aileron deflection causes a negative rolling moment. Cnδa is, however, positive. So a positive aileron deflection causes positive yaw. 7.4.2 Rudder deflections The rudder stability derivatives are CYδr , Clδr and Cnδr . The forces and moments caused by a rudder deflection are given by 1 Y = CYδr δr ρV 2 S, 2 1 L = Clδr δr ρV 2 Sb. 2 The coefficient CYδr is given by CYδr = CYvδ The coefficient Clδr is then given by Clδr = CYδr   Vv V 2 and 1 N = Cnδr δr ρV 2 Sb. 2 Sv . S  xv − xcg zv − zcg cos α0 − sin α0 . b b (7.4.3) (7.4.4) (7.4.5) Clδr is positive. This means that a positive rudder deflection causes a positive rolling moment. This effect is generally not desirable. Especially if zv − zcg is big, measures are often taken to reduce this effect. Finally, the rudder effectiveness Cnδr is given by Cnδr = −CYδr lv . b (7.4.6) This coefficient is negative. A positive rudder deflection thus causes a negative yawing moment. 7.4.3 Spoiler deflections The last things we examine are the spoilers. Spoilers are often used in high-speed aircraft to provide roll control. A spoiler deflection δs on the left wing is defined to be positive. Due to this definition, we have Clδs < 0 and Cnδs < 0. Of these two, the latter is the most important. 34 8. Longitudinal stability and control In this chapter, we will start to investigate the stability of the entire aircraft. This can be split up into two parts: longitudinal and lateral stability. In this chapter, we will only look at longitudinal stability. 8.1 8.1.1 Stick fixed longitudinal stability Effects of the wing and the tail on stability To start our investigation in the stability of an aircraft, we reexamine the moment equation. In an earlier chapter, we found that Cm = Cmac + CNwα (α − α0 ) xcg − xw − CNh c̄  Vh V 2 Sh l h = 0, Sc̄ (8.1.1) where CNh is given by CNh = CNhα     dε + (α0 + ih ) + CNhδ δe . (α − α0 ) 1 − e dα (8.1.2) We can also rewrite the moment equation to Cm = Cmw + Cmh . In this equation, Cmw is the contribution due to the wings. Similarly, Cmh is the contribution from the horizontal tailplane. They are both given by  2 Sh l h Vh xcg − xw and Cmh = −CNh . (8.1.3) Cmw = Cmac + CNwα (α − α0 ) c̄ V Sc̄ Taking a derivative of the moment equation will give us Cmα = Cmαw + Cmαh , where Cmαw xcg − xw = CNwα c̄ and Cmαh = −CNhα  dε 1− dα  Vh V 2 Sh l h . Sc̄ (8.1.4) To achieve stability for our aircraft, we should have Cmα < 0. Usually, the wing is in front of the CG. We thus have xcg − xw > 0 and also Cmαw > 0. The wing thus destabilizes the aircraft. Luckily, the horizontal tailplane has a stabilizing effect. This is because Cmαh < 0. To achieve stability, the stabilizing effect of the tailplane should be bigger than the destabilizing effect of the wings. We should thus have |Cmαw | < |Cmαh |. 8.1.2 (8.1.5) Effects of the center of gravity on stability We will now examine the effects of the CG on the stability. To do this, we suppose xcg increases (the CG moves to the rear). However, the other parameters (including δe ) stay constant. The movement of the CG causes Cmα to increase. At a certain point, we will reach Cmα = 0. When the CG moves beyond this position, the aircraft becomes unstable. Let’s examine the point at which Cmα = 0. We remember, from a previous chapter, that this point is called the neutral point. And, because the stick deflection is constant (δe is constant), we call this point the stick fixed neutral point. Its x coordinate is denoted by xnf ix . To find it, we can use Cmα = CNwα xnf ix − xw + CNhα c̄  35 1− dε dα  Vh V 2 Sh xnf ix − xh . S c̄ (8.1.6) After some mathematical trickery, we can find the position of the stick fixed neutral point, with respect to the wing. It is given by CNhα xnf ix − xw = c̄ CNα  dε 1− dα  Vh V 2 Sh l h . Sc̄ (8.1.7) From this, we can also derive the position of the stick fixed neutral point, with respect to the aircraft CG. This is given by xcg − xnf ix . (8.1.8) Cmα = CNα c̄ xcg −xn f ix is known as the (stick fixed) stability margin. It is an indication of how much The quantity c̄ the CG can move, before the aircraft becomes unstable. 8.1.3 The elevator trim curve Now let’s examine the effects of the elvator deflection δe . We know from a previous chapter that the elevator deflection necessary to keep the aircraft in equilibrium is δe = − 1 (Cm0 + Cmα (α − α0 )) . Cmδe (8.1.9) δe depends on α. To see how, we plot δe versus α. We usually do this, such that the y axis is reversed. (Positive δe appear below the horizontal axis.) Now we examine the slope of this graph. It is given by dδe Cmα =− . dα Cmδe (8.1.10) We always have Cmδe < 0. To make sure we have Cmα < 0 as well, we should have dδe /dα < 0. The line in the δe , α graph should thus go upward as α increases. (Remember that we have reversed the y axis of the graph!) δe also depends on the aircraft velocity V . To see how, we will rewrite equation (8.1.9). By using W CN ≈ CNα (α − α0 ) ≈ 1 ρV 2 S , we find that 2 δe = − 1 Cmδe  Cm0 Cmα W + CNα 12 ρV 2 S  . (8.1.11) We can now plot δe against V . (Again, we reverse the δe axis.) We have then created the so-called elevator trim curve. Its slope is given by 4W 1 Cmα dδe . = dV ρV 3 S Cmδe CNα (8.1.12) To have Cmα < 0, we should have dδe /dV > 0. The line in the graph should thus go downward. Also, if you want to fly faster in a stable aircraft, you should push your stick forward. 8.2 8.2.1 Stick free longitudinal stability The stick free elevator deflection Previously, we have assumed that δe is constant. The pilot has his stick fixed. But what will happen if the pilot releases his stick? It would be nice if the aircraft remains stable as well. 36 Let’s suppose the pilot releases the stick. In that case, aerodynamic force will give the elevator a certain stick free elevator deflection δef ree . To find δef ree , we examine the moments He about the elevator hinge point. (Or, to be more precise, we look at the non-dimensional version Che .) Contributing to this hinge moment are the horizontal tailplane, the elevator and the trim tab. By using a linearization, we find that Chef ree = Chα αh + Chδ δef ree + Chδt δte = 0. (8.2.1) It follows that the stick free elevator deflection is δef ree = − Chδt Chα αh − δt . Chδ Chδ e   (8.2.2) From this, we can also derive that  dδe dα f ree Ch =− α Chδ dε 1− dα  . (8.2.3) The elevator deflection thus changes as the angle of attack is changed. 8.2.2 Differences in the moment due to the stick free evelator The free elevator deflection effects the contribution Cmh of the horizontal tailplane to the moment Cm . Let’s investigate this. We can remember that    V 2 S l h h h Cmh = − CNhα αh + CNhδ δe . V Sc̄ (8.2.4) We now substitute δe by δef ree . If we also differentiate with respect to α, and work things out, we will get    2  dε Chα Vh Sh l h 1− . (8.2.5) Cmαh = − CNhα − CNhδ f ree Chδ dα V Sc̄ If we compare this equation to the right side of equation (8.1.4), we see that only CNhα has changed. In fact, we can define Ch (8.2.6) = CNhα − CNhδ α . CNhα f ree Chδ If we use CNhα f ree , instead of CNhα , then our stability analysis is still entirely valid. Let’s take a closer look at the differences between CNhα f ree and CNhα . This difference is the term Chα Chδ CNhδ . We know that CNhδ > 0. The term Chδ is interesting. If it would be positive, then it can be shown that the elevator position is unstable. So, we have to have Chδ < 0. Finally there is Chα . This term can be either positive or negative. If it is positive (Chα > 0), then the stick free aircraft will be more stable than the stick fixed aircraft. If, however, it is negative (Chα < 0), then it will be less stable, or possibly even unstable. 8.2.3 The stick free neutral point Let’s find the stick free neutral point xnf ree . Finding xnf ree goes similar to finding xnf ix . In fact, we can adjust equations (8.1.7) and (8.1.8) to   2  CNhα xnf ree − xw Vh Sh l h dε f ree 1− = , c̄ CNα dα V Sc̄ 37 (8.2.7) xcg − xnf ree . (8.2.8) c̄ In this equation, we have CNαf ree ≈ CNα . This is because the elevator has a negligible influence on CNα , compared to the influence of the wing. Cmαf ree = CNαf ree We can also find the position of the stick free neutral point, with respect to the stick fixed neutral point. Subtracting equation (8.1.7) from equation (8.2.7) gives   2    CNhδ Chα xnf ree − xnf ix Vh Sh l h Cmδ Chα dε dε . (8.2.9) =− = 1− 1− c̄ CNα Chδ dα V Sc̄ CNα Chδ dα 8.2.4 Elevator stick forces Now we will examine the stick forces which the pilot should exert. We denote the stick deflection by se . By considering the work done by the pilot, we find that Fe dse + He dδe = 0. From this follows that the stick force Fe is given by Fe = − 1 dδe dδe He = − Ch ρV 2 Se c̄e . dse dse e 2 h (8.2.10) By the way Se is the elevator surface and c̄e is the mean elevator chord. If we massively rewrite the above equation, we can eventually find that   2  Vh 1 dδe W 1 . (8.2.11) Ch′ 0 ρV 2 + Ch′ α Se c̄e Fe = − dse V 2 S CNα We see that Fe consists of two parts. One part varies with the airspeed, while the other part does not. By the way, the coefficients Ch′ 0 and Ch′ α are given by Chδ Chδ Cmac − (α0 + ih ) + Chδt δte , CNhα f ree Cmδ CNhδ xcg − xnf ree Chδ Ch =− Cmαf ree = − δ CNα . Cmδ Cmδ c̄ Ch′ 0 = − (8.2.12) Ch′ α (8.2.13) We see that Ch′ 0 depends on δte . To simplify our equation, we can apply a small trick. We define δte0 to be the value of δte for which Ch′ 0 = 0. It follows that ! 1 Chδ Chδ (α0 + ih ) . (8.2.14) δte0 = Cmac + CNhα f ree Chδt Cmδ CNhδ We can now rewrite the stick deflection force as  2  
W Chδ xcg − xnf ree Vh 1 2 dδe Se c̄e − ρV Chδt δte − δte0 . Fe = dse V S Cmδe c̄ 2 (8.2.15) The control forces, which the pilots need to exert, greatly determine how easy and comfortable it is to fly an airplane. The above equation is therefore rather important. We can also derive something else from the above equation. Let’s define the trim speed Vtr to be the speed at which Fe = 0. We now examine the derivative dFe /dV at this trim speed. (So at Fe = 0.) If it is positive (dFe /dV > 0), then the aircraft is said to have elevator control force stability in the current flight condition. It can be shown that this derivative is given by  2   W Chδ xcg − xnf ree 1 Vh dδe dFe = −2 Se c̄e . (8.2.16) dV Fe =0 dse Vtr S Cmδe c̄ Vtr It’s the job of the designer to keep this derivative positive. 38 8.3 8.3.1 Longitudinal control Special manoeuvres Previously, we have only considered steady flight. Now we suppose that we are performing some special manoeuvre. We will consider both a steady pull-up manoeuvre and a horizontal steady turn. During these manoeuvres, we will have a certain load factor n = N/W . There are two parameters that are important for the manoeuvres. They are the elevator deflection per g, denoted by dδe /dn, and the stick force per g, denoted by dFe /dn. Both these parameters should be negative. And they may not be too high or too low either. 8.3.2 The elevator deflection per g We will now find an expression for dδe /dn. Let’s suppose we’re initially in a horizontal steady flight. But after a brief moment, we’ll be in one of the special manoeuvres. In this brief moment, several aircraft parameters have changed. Let’s examine the change in normal force ∆CN and the change in moment ∆Cm . The change in normal force is effected by the angle of attack α and the pitch rate q. This gives us ∆CN = ∆N 1 2 2 ρV S = W ∆n 1 2 2 ρV S = CNα ∆α − CZq ∆ qc̄ . V (8.3.1) Similarly, the change in moment is effected by the angle of attack α, the pitch rate q and the elevator deflection δe . This gives us qc̄ (8.3.2) ∆Cm = 0 = Cmα ∆α + Cmq ∆ + Cmδe ∆δe . V You may wonder, why is ∆Cm = 0? This is because both in the initial situation and the final situation, we have a steady manoeuvre. There is thus no angular acceleration present. The moment must thus stay constant. From the first of the above two equations, we can find the derivative of α with respect to n. It is given by CZq d qc̄ W 1 dα V = . (8.3.3) + dn CNα 12 ρV 2 S CNα dn From the second of these equations, we can find that dδe 1 =− dn Cmδe qc̄ d dα Cmα + Cmq V dn dn ! . (8.3.4) Inserting the value of dα/dn will eventually give us 1 dδe =− dn Cmδe Cmα W + CNα 12 ρV 2 S  Cmα CZq + Cmq CNα  d qc̄ V dn ! . (8.3.5) We will determine the term d qc̄ V /dn later, since it depends on the type of manoeuvre that is being performed. 8.3.3 The stick force per g It’s time to find an expression for dFe /dn. From equation (8.2.10), we can derive that   dδe dδe 1 2 dαh dFe . =− ρV Se c̄e Chα + Chδ dn dse 2 h dn dn 39 (8.3.6) We already have an expression for dδe /dn. The expression for αh is a bit tricky. This is because we also have a rotation q. If we take this into account, we will have   lh qc̄ dε + (α0 + ih ) + . (8.3.7) αh = (α − α0 ) 1 − dα c̄ V The derivative of αh , with respect to n, will then be   qc̄ dε dα lh d V dαh = 1− + . dn dα dn c̄ dn (8.3.8) Luckily, we still remember dα/dn from equation (8.3.3). From this, we can derive an equation that’s way too long to write down here. However, once we examine specific manoeuvres, we will mention the final equation. 8.3.4 The pull-up manoeuvre Let’s consider an aircraft in a pull-up manoeuvre. When an aircraft pulls its nose up, the pilot will experience higher g-forces. This will thus cause the load factor n to change. To be able to study pull-up manoeuvres, we simplify them. We assume that both n and V are constant. If this is the case, the aircraft’s path will form a part of a circle. The centripetal accelaration thus is N − W = mV q. By using n = N/W and W = mg, we can rewrite this as gc̄ qc̄ = 2 (n − 1). V V (8.3.9) Differentiating with respect to n gives d qc̄ W gc̄ 1 V , = 2 = dn V 2µc 12 ρV 2 S where µc = m W = . ρSc̄ gρSc̄ (8.3.10) By using this, we can find the elevator deflection per g for a pull-up manoeuvre. It is     CZq Cmq Cmα W 1 dδe 1 + =− + . dn Cmδe 12 ρV 2 S CNα 2µc 2µc (8.3.11) Often the term CZq /2µc can be neglected. This simplifies matters a bit. We can also derive a new expression for the stick force per g. We will find that !  2 Cmqf ree Cmαf ree Chδ dδe W Vh dFe . (8.3.12) Se c̄e + = dn dse S V Cmδe CNα 2µc In this equation, we can see the parameters Cmαf ree and Cmqf ree . These are the values of Cmα and Cmq when the pilot releases his stick. They are given by Cmαf ree = CNwα (The relation for Cmαh xcg − xw + Cmαh f ree c̄ f ree 8.3.5 and Cmqf ree = Cmq − Cmδe Chα lh . CNα c̄ (8.3.13) was already given in equation (8.2.5).) The steady horizontal turn Now let’s consider an aircraft in a steady horizontal turn. It is performing this turn with a constant roll angle ϕ. From this, we can derive that N cos ϕ = W and N − W cos ϕ = mV q. 40 (8.3.14) If we combine the above relations, and rewrite them, we will get   1 qc̄ gc̄ . = 2 n− V V n (8.3.15) Differentiating with respect to n will then give us d qc̄ W 1 V = 1 dn 2µc 2 ρV 2 S  1+ 1 n2  . By using this, we can find the elevator deflection per g for a horizontal steady turn. It is     Cmq dδe Cmα 1 W 1 Cmα CZq 1 + . + + =− dn Cmδe 12 ρV 2 S CNα CNα 2µc 2µc n2 (8.3.16) (8.3.17) Again, we may often assume that CZq /2µc ≈ 0. This again simplifies the equation. We also have the stick force per g. In this case, it is given by !  2 Cmqf ree  Cmαf ree 1 dδe W Vh Chδ dFe 1+ 2 . (8.3.18) + = Se c̄e dn dse S V Cmδe CNα 2µc n It is interesting to see the similarities between the pull-up manoeuvre and the steady horizontal turn. In fact, if the load factor n becomes big, the difference between the two manoeuvres disappears. 8.3.6 The manoeuvre point An important point on the aircraft, when performing manoeuvres, is the manoeuvre point. It is defined as the position of the CG for which dδe /dn = 0. First we will examine the stick fixed manoeuvre point xmf ix . To have dδe /dn = 0 for a pull-up manoeuvre (neglecting CZq /2µc ), we should have xcg − xnf ix Cmq Cmq Cmα + = = 0. + CNα 2µc c̄ 2µc (8.3.19) If the above equation holds, then the CG equals the manoeuvre point. We thus have xmf ix − xnf ix Cmq =− c̄ 2µc and also xcg − xmf ix Cmq Cmα . + = c̄ CNα 2µc (8.3.20) (Remember that the above equations are for the pull-up manoeuvre. For the steady turn, we need to
multiply the term with Cmq by an additional factor 1 + 1/n2 .) By using the above results, we can eventually obtain that dδe W xcg − xmf ix 1 =− . (8.3.21) 1 dn Cmδe 2 ρV 2 S c̄ By the way, this last equation is valid for both the pull-up manoeuvre and the steady horizontal turn. We can also find the stick free manoeuvre point xmf ree . This goes, in fact, in a rather similar way. We will thus also find, for the pull-up manoeuvre, that Cmqf ree Cmαf ree xcg − xmf ree . + = c̄ CNα 2µc
(For the steady turn, we again need to multiply the term with Cmqf ree by 1 + 1/n2 .) Cmqf ree xmf ree − xnf ree =− c̄ 2µc and 41 (8.3.22) 9. Lateral stability and control In this chapter, we will examine lateral stability and control. How should we control an aircraft in a non-symmetrical steady flight? 9.1 9.1.1 The equations of motion Derivation of the equations of motion for asymmetric flight Let’s examine an aircraft in a steady asymmetric flight. It has a roll angle ϕ and a sideslip angle β. By examining equilibrium, we can find that W sin ϕ + Y = mV r, L=0 and N = 0. (9.1.1) Non-dimensionalizing these equations gives rb + CY = 0, Cl = 0 and Cn = 0, (9.1.2) 2V m . We can also apply linearization to the above equations. This will then give us where we have µb = ρSb CL ϕ − 4µb rb + CYδa δa + CYδr δr = 0, 2V rb + Clδa δa + Clδr δr = 0, Clβ β + Clr 2V rb Cnβ β + Cnr + Cnδa δa + Cnδr δr = 0. 2V CL ϕ + CYβ β + (CYr − 4µb ) 9.1.2 (9.1.3) (9.1.4) (9.1.5) Simplifying the equations of motion Let’s examine the equations of the previous paragraph. There are quite some terms in these equations that are negligible. They are CYδa , Clδr , CYr , Cnδa and CYδr . By using these neglections, and by putting the above equations into matrix form, we will get   ϕ      β 0 CL CYβ −4µb 0 0     rb  = (9.1.6) 0 . 0  Clβ Clr Clδa   0  2V    0 0 Cnδr  δa  0 Cnβ Cnr δr Let’s assume that the velocity V is already set. We then still have five unknowns and three equations. That means that there are infinitely many solutions. This makes sense: You can make a turn in infinitely many ways. How do we deal with this? We simply set one of the parameters. We then express three of rb .) So the remaining parameters as a function of the last parameter. (This fifth parameter is usually 2V let’s do that. 9.2 9.2.1 Steady horizontal turns Turns using ailerons only Let’s try to turn the aircraft, by only using ailerons. We do not use the rudder and thus have δr = 0. We can insert this into the equations of motion. We then solve for the parameters β, ϕ and δa . This will 42 give us dβ Cn =− r >0 rb Cnβ d 2V (since Cnr < 0 and Cnβ > 0), (9.2.1) C 4µb + CYβ Cnnr dϕ β = > 0, rb CL d 2V (9.2.2) dδa 1 Clβ Cnr − Clr Cnβ . = rb Clδa Cnβ d 2V (9.2.3) rb The sign of the last equation is still a point of discussion. We would like to have dδa /d 2V . If this is the case, then we have so-called spiral stability. We know that Clδa < 0 and Cnβ > 0. So spiral stability is achieved if (9.2.4) Clβ Cnr − Clr Cnβ > 0. We will find out in the next chapter why they call this the spiral stability condition. 9.2.2 Turns using the rudder only We can also make a turn using only the rudder. So we have δa = 0. This again gives us three equations, being dβ Cl =− r >0 (9.2.5) (since Clr > 0 and Clβ < 0), rb Clβ d 2V C 4µb + CYβ Cllr dϕ β > 0, = rb CL d 2V (9.2.6) dδr 1 Clβ Cnr − Clr Cnβ . =− rb Cnδr Clβ d 2V (9.2.7) In the last equation, we have Cnδr < 0 and Clβ < 0. If there is also spiral stability, then we have rb dδr /d 2V < 0. 9.2.3 Coordinated turns In a coordinated turn, we have β = 0. This means that there is no sideward component of the force acting on the aircraft. This is an important factor for passenger comfort. For the coordinated turn, we again have three equations. They are dϕ 4µb > 0, (9.2.8) = rb CL d 2V dδa Cl =− r >0 rb Clδa d 2V (since Clr > 0 and Clδa < 0), (9.2.9) Cnr dδr =− <0 rb Cnδr d 2V (since Cnr < 0 and Cnδr < 0). (9.2.10) 43 9.2.4 Flat turns If we want the aircraft to stay flat during the turns, then we have ϕ = 0. It then follows that dβ 4µb < 0. = rb CYβ d 2V (9.2.11) From this, we can also derive that dδa >0 rb d 2V 9.3 9.3.1 and dδr < 0. rb d 2V (9.2.12) Other flight types Steady straight sideslipping flight Let’s examine a steady straight sideslipping flight. This type of flight is usually only used during landings with strong sidewinds. However, sometimes the aircraft is brought into a steady straight sideslipping flight involuntarily. It is therefore important to know how the aircraft behaves. In a straight flight, we have rb 2V = 0. We can now derive that CY dϕ = − β > 0, dβ CL Cl dδa =− β dβ Clδa and Cn dδr =− β . dβ Cnδr (9.3.1) We generally want to have dδa /dβ < 0 and dδr /dβ > 0. We also always have Clδa < 0 and Cnδr < 0. This implies that we should have Clβ < 0 and Cnβ > 0. 9.3.2 Stationary flight with asymmetric power Let’s suppose one of the engines of the aircraft doesn’t work anymore. In this case, a yawing moment will be present. This moment has magnitude ∆Tp ye Cne = k 1 2 . 2 ρV Sb (9.3.2) The variable ∆Tp consists of two parts. First there is the reduction in thrust. Then there is also the increase in drag of the malfunctioning engine. ye is the Y coordinate of the malfunctioning engine. Finally, k is an additional parameter, taking into account other effects. Its value is usually between 1.5 and 2. Now let’s try to find a way in which we can still perform a steady straight flight. (We should thus have r = 0.) We now have four unknowns and three equations. So we can still set one parameter. Usually, we would like to have ϕ = 0 as well. In this case, a sideslip angle β is unavoidable. If the right engine is inoperable, then a positive rudder deflection and sideslip angle will be present. We could also choose to have β = 0. In this case, we will constantly fly with a roll angle ϕ. The wing with the inoperable engine then has to be lower than the other wing. So if the right wing malfunctions, then we have a positive roll angle. 44 10. Aircraft modes of vibration It is finally time to look at the dynamics of an aircraft. How will an aircraft behave, when given elevator, rudder and aileron deflections? What are its modes of vibration? That’s what we will look at in this chapter. 10.1 Eigenvalue theory 10.1.1 Solving the system of equations To examine the dynamic stability of the aircraft, we examine the full longitudinal equations of motion. The symmetric part of these equations were      0 CXq CZ0 CXα CXu − 2µc Dc û       CZu CZα + (CZα̇ − 2µc )Dc −CX0 2µc + CZq    α  0 (10.1.1)    =  .    θ  0 0 0 −Dc 1 qc̄ 0 0 Cmq − 2µc KY2 Dc Cmα + Cmα̇ Dc Cmu V In this system of equations, we have assumed stick-fixed conditions. All inputs are zero. We could try to find solutions for the above system of equations. A common way to do this, is to assume a solution of the form x(t) = Aeλc sc . (10.1.2) In this equation, x(t) is our solution. sc = Vc̄ t is the dimensionless time. From this form follows that Dc x = λc x. If we insert this into the equations of motion, we find      CXq CZ0 CXα CXu − 2µc λc Au 0      2µc + CZq  Aα  λc sc 0 CZα + (CZα̇ − 2µc )λc −CX0 CZu   (10.1.3) =  .   e    Aθ  0 0 0 −λc 1 0 Cmu Cmα + Cmα̇ λc 0 Cmq − 2µc KY2 λc Aq We can write this matrix equation as [∆] Aeλc sc = 0. The exponential in this equation can’t be zero, so we can get rid of it. We thus need to solve [∆] A = 0. One solution of this equation is A = 0. However, this is a rather trivial solution, in which we are not interested. So we need to find non-trivial solutions. This is where our knowledge on linear algebra comes in. There can only be non-trivial solutions, if det [∆] = 0. Applying this will give us an equation of the form Aλ4c + Bλ3c + Cλ2c + Dλc + E = 0. (10.1.4) This equation is called the characteristic polynomial. Solving it will give four eigenvalues λc1 , λc2 , λc3 and λc4 . Corresponding to these four eigenvalues are four eigenvectors A1 , A2 , A3 and A4 . The final solution of the system of equations now is x = c1 A1 eλc1 sc + c2 A2 eλc2 sc + c3 A3 eλc3 sc + c4 A4 eλc4 sc . (10.1.5) The constants c1 , c2 , c3 and c4 depend on the four initial conditions. 10.1.2 The eigenvalues Let’s examine the eigenvalues of the system of equations. Each eigenvalue can be either real and complex. If one of the eigenvalues is complex, then its complex conjugate is also an eigenvalue. Complex eigenvalues therefore always come in pairs. 45 A mode of vibration is a characteristic way in which an object (our aircraft) can vibrate. The number of modes depends on the eigenvalues. In fact, it is equal to the number of different eigenvalues. (When performing the counting, a pair of complex conjugate eigenvalues is counted as one.) For example, an aircraft with two real eigenvalues and two complex eigenvalues has three modes of vibration. The eigenvalues λc are very important for the stability of the system. To examine stability, we look at the limit lim c1 A1 eλc1 sc + c2 A2 eλc2 sc + c3 A3 eλc3 sc + c4 A4 eλc4 sc . (10.1.6) sc →∞ If only one of the eigenvalues has a positive real part, then this limit will diverge. This means that our aircraft is unstable. If, however, all eigenvalues have negative real parts, then the system is stable. 10.1.3 Real eigenvalue properties We can derive some interesting properties from the eigenvalues. First, let’s examine a real eigenvalue λc . This eigenvalue has its own mode of vibration x = Aeλc sc . The half time T 21 is defined as the time it takes to reduce the amplitude of the motion to half of its former magnitude. In other words, x(t + T 12 ) = 21 x(t). Solving this equation will give T 21 = ln 21 c̄ . λc V (10.1.7) Similarly, we define the time constant τ as the time it takes for the amplitude to become 1/e of its former magnitude. Solving x(t + τ ) = 1e x(t) gives τ =− 1 c̄ . λc V (10.1.8) These two parameters of course only exist if λc is negative. If it is positive, then the magnitude will only grow. In this case, the doubling time T2 is an important parameter. It is given by T2 = −T 21 . 10.1.4 Complex eigenvalue properties √ Now let’s examine a complex eigenvalue pair. We can write it as λc1,2 = ξc ± ηc i, where i = −1 is the complex number. This eigenvalue will cause an oscillation. The period and frequency of the oscillation only depend on ηc . In fact, the period P , the frequency f and the angular frequency ωn are given by 2π c̄ 1 ηc V 2π V P = , f= = and ωn = = ηc . (10.1.9) ηc V P 2π c̄ P c̄ The damping of this oscillation is caused by the real part ξc . Again, the half time T 21 is defined as the time it takes for the amplitude to reduce to half its size. It is still given by T 12 = ln 21 c̄ . ξc V (10.1.10) Another important parameter is the logarithmic decrement δ. It is defined as the natural logarithm of the ratio of the magnitude of two successive peaks. In other words, it is defined as ! V V eξc c̄ (t+P ) = ξc P. δ = ln (10.1.11) V t ξ c c̄ e c̄ Finally, there are the damping ratio ζ and the undamped angular frequency ω0 . They are defined such that  c̄  p . (10.1.12) λc1,2 = −ζω0 ± iω0 1 − ζ 2 V 46 Solving for ζ and ω0 will give −ξc ζ=p ξc2 + ηc2 10.1.5 and ω0 = p V ξc2 + ηc2 . c̄ (10.1.13) Getting stable eigenvalues Let’s take a look at the characteristic equation (equation (10.1.4)). We usually set up the equation, such that A > 0. To obtain four eigenvalues with negative real parts, we must have B > 0, C > 0, D > 0 and E > 0. (10.1.14) But these aren’t the only conditions to ensure that we have stable eigenvalues. We must also have R = BCD − AD2 − B 2 E > 0. (10.1.15) These criteria are known as the Routh-Hurwitz Stability Criteria. The coefficient R is called Routh’s discriminant. These criteria hold for both the symmetric and the asymmetric modes of vibration. 10.2 The symmetric modes of vibration 10.2.1 Example eigenvalues Let’s suppose we know all the parameters in the matrix equation that was described earlier. In this case, we can find the four eigenvalues. An example solution of these eigenvalues is given by λc1,2 = −0.04 ± 0.04i and λc3,4 = −0.0003 ± 0.006i. (10.2.1) Of course these values will be different for different aircraft. But most types of aircraft, having the standard wing-fuselage-tailplane set-up, will have similar eigenvalues. Let’s study these eigenvalues. There are two pairs of complex conjugate eigenvalues. Both pairs of eigenvalues have negative real parts. This means that the aircraft is stable. Since there are only two pairs of complex conjugate eigenvalues, there are two modes of vibration. We will now examine these modes. 10.2.2 The short period oscillation Let’s look at the first pair of eigenvalues. It has a relatively big real part ξc The damping is therefore big. The complex part ηc is relatively big as well. So the frequency is high. In other words, we have a highly damped high-frequency oscillation. This motion is known as the short period oscillation. Let’s take a look at what actually happens in the aircraft. We start the short period oscillation by applying a step input to the elevator deflection. (We deflect it, and keep that deflection.) We can, for example, deflect it upward. This causes the lift on the horizontal tailplane to decrease. This, in turn, causes the pitch rate to increase. An increase in pitch rate will, however, increase the effective angle of attack of the horizontal tailplane. This then reduces the pitch rate. And the whole cycle starts over again. However, the oscillation is highly damped. After less than one period, the effects are hardly noticable anymore. Now let’s try to derive some equations for the short period motion. The short period motion is rather fast. So we assume the aircraft hasn’t had time yet to change its velocity in X or Z direction. This 47 means that û = 0 and γ = 0. Therefore α = θ. This reduces the equations of motion to #" # " # " 2µc + CZq α CZα + (CZα̇ − 2µc )λc 0 . qc̄ = 2 Cmq − 2µc KY λc Cmα + Cmα̇ λc 0 V (10.2.2) We can now find the eigenvalues for this matrix. This will still give us a rather complicated equation. If we neglect CZα̇ and CZq , then this complicated equation reduces to λc1,2 √ −B ± i 4AC − B 2 . = 2A (10.2.3) In this equation, the coefficients A, B and C are given by A = 4µ2c KY2 , B = −2µc (KY2 CZα + Cmα̇ + Cmq ) and C = CZα Cmq − 2µc Cmα . (10.2.4) B negative. We know that A is positive. This means that B has to To have stability, we should have − 2A be positive as well. 10.2.3 The phugoid Now we look at the second pair of eigenvalues. It has a small real part ξc , and therefore a small damping. The complex part ηc is small as well, so the frequency is low. In other words, we have a lightly damped low-frequency oscillation. This motion is known as the phugoid. Again, we look at what happens with the aircraft. This time, we apply an impulse deflection on the elevator. (We only deflect it briefly.) This will cause our pitch angle to increase. (That is, after the short period motion has more or less damped out.) We will therefore go upward. This causes our velocity to decrease. Because of this, the lift is reduced. Slowly, the pitch angle will decrease again, and we will go downward. This causes the velocity to increase. This, in turn, increases the lift. The pitch angle will again increase, and we will again go upward. Again, we will try to derive some relations for the phugoid. In the phugoid, the angle of attack α is approximately constant. (γ and θ do vary a lot though.) So we have α = 0 and α̇ = 0. (Remember that we’re discussing deviations from the initial position.) Since the oscillation is very slow, we also assume that q̇ = 0. If we also neglect the terms CZq and CX0 , we will find that we again have λc3,4 = √ −B ± i 4AC − B 2 . 2A (10.2.5) However, now the coefficients are given by A = −4µ2c , B = 2µc CXu and C = −CZu CZ0 . (10.2.6) We can apply the approximations CXu = −2CD , CZ0 = −CL and CZu = −2CL . This would then give us the three parameters s √ CL2 2 CD 2π √ V V 2π g√ ≈ ω0 = 2, ζ = and P = p = 2π . = (10.2.7) 2 c̄ 2µc V 2 CL ω0 g ω0 1 − ζ 2 Note that, in the above equation for P , we have used the fact that the damping ζ is small. Although the above equations are only approximations, they can serve as quite handy tools in verifying your results. 48 10.3 The asymmetric modes of vibration 10.3.1 Example eigenvalues We have just examined the symmetric equations of motion. Of course, we can do the same for the asymmetric equations of motion. These equations of motion are      CL CYp CYr − 4µb CYβ + (CYβ̇ − 2µb )Db 0 β    ϕ  0 1 D 1 0 0 −      2 b (10.3.1)    pb  =   . 2  0  Clβ 0 Clp − 4µb KX Db Clr + 4µb KXZ Db   2V rb 0 0 Cnp + 4µb KXZ Db Cnr − 4µb KZ2 Db Cnβ + Cnβ̇ Db 2V Examining them goes in more or less the same way as for the symmetric case. There is, however, one important difference. Since we are examining the asymmetric case, we don’t use the chord c̄ but we use the wing span b. The eigenvalues are thus also denoted as λb . Example eigenvalues for an aircraft are λb1 = −0.4, λb2 = 0.01 and λb3,4 = −0.04 ± 0.4i. (10.3.2) You might be surprised that these eigenvalues are a lot bigger than the symmetric eigenvalues. This is not very important. It’s only the case, because they are based on b, instead of on c̄. And naturally, b is a lot bigger than c̄. Let’s examine the eigenvalues. There are two real eigenvalues and one pair of complex conjugate eigenvalues. The aircraft thus has three modes of vibration. You might also have noticed that there is a positive eigenvalue. The aircraft is thus unstable. The eigenvalue is, however, very small. This means that the aircraft state will only diverge very slowly. The pilot will have plenty of time to do something about it. So you don’t have to worry: flying is still safe. 10.3.2 The aperiodic roll The motion corresponding to λb1 is called the aperiodic roll. The eigenvalue is very negative. This motion is therefore highly damped. The aperiodic roll is induced by applying a step input to the aileron. When this happens, the aircraft will start rolling. Let’s suppose it rolls to the right. The right wing then goes down. This means that the right wing will get a higher effective angle of attack. The lift of this wing thus increases. The opposite happens for the left wing: its lift decreases. This lift difference causes a moment opposite to the rolling motion. In other words, the motion is damped out. The roll rate p will converge rather quickly to a constant value. The aperiodic roll is a very fast motion. So there is no time for sideslip or yaw effects to appear. So we can assume that, during an aperiodic roll motion, we have β = r = 0. This reduces the equations of motion to just one equation, being
pb = 0. 2V It directly follows that the corresponding eigenvalue is given by 2 Db Clp − 4µb KX λb1 = 10.3.3 Clp 2 . 4µb KX (10.3.3) (10.3.4) The spiral motion The motion corresponding to λb2 is called the spiral motion. The eigenvalue is positive. So the motion is unstable. However, the eigenvalue is very small. This means that divergence will occur only very 49 slowly. We thus say that the motion is marginally unstable. (For some aircraft, this value is slightly negative. Such aircraft are marginally stable.) The spiral motion is induced by an initial roll angle. (An angle of 10◦ is sufficient.) This causes the lift vector to be tilted. The horizontal component of the lift will cause the aircraft to make a turn. In the meanwhile, the vertical component of the lift vector has slightly decreased. This causes the aircraft to lose altitude. Combining these two facts will mean that the aircraft will perform a spiral motion. If the eigenvalue λb2 is positive, then the roll angle of the aircraft will slowly increase. The spiral motion will therefore get worse. After a couple of minutes, the roll angle might have increased to 50◦ . This phenomenon is, however, not dangerous. The pilot will have plenty of time to react. It is also very easy to pull the aircraft out of a spiral motion. Let’s try to derive an equation for λb2 . The spiral motion is a very slow motion. We thus neglect the derivatives of β, p and r. Also, the coefficients CYr and CYp are neglected. After working out some equations, we can eventually find that
2CL Clβ Cnr − Cnβ Clr
.
(10.3.5) λ b2 = Clp CYβ Cnr + 4µb Cnβ − Cnp CYβ Clr + 4µb Clβ The denominator of this relation is usually negative. We say we have spiral stability if λb2 < 0. This is thus the case if (10.3.6) Clβ Cnr − Cnβ Clr > 0. You might remember that we’ve seen this equation before. 10.3.4 The Dutch roll The pair of eigenvalues λb3,4 has a slightly low damping and a slightly high frequency. In the mode of vibration corresponding to these eigenvalues, the aircraft alternately performs a yawing and a rolling motion. The mode of vibration is called the Dutch roll. Let’s take a look at what actually happens with the aircraft. To initiate the Dutch roll, an impulse input is applied to the rudder. This causes the aircraft to yaw. Let’s suppose the aircraft yaws to the right. The lift on the left wing then increases, while the lift on the right wing decreases. This moment causes the aircraft to roll to the right. When the aircraft is rolling to the right, then the lift vector of the right wing is tilted forward. Similarly, the left wing will have a lift vector that is tilted backward. This causes the aircraft to yaw to the left. (This effect is still called adverse yaw.) In this way, roll and yaw alternate each other. It is important to remember that roll and yaw are alternately present. When the roll rate is at a maximum, the yaw rate is approximately zero, and vice verse. The Dutch roll is not very comfortable for passenger. To increase passenger comfort, a yaw damper is used. This is an automatic system, which uses rudder/aileron deflections to reduce the effects of the Dutch roll. Let’s try to find a relation for λb3,4 . This is rather hard, since both roll and yaw are present. However, experience has shown that we still get slightly accurate results, if we neglect the rolling part of the motion. We thus assume that ϕ = p = 0. This reduces the system of equations to a 2 × 2 matrix. From it, we can again find that √ −B ± i 4AC − B 2 . (10.3.7) λb3,4 = 2A However, this time the coefficients A, B and C are given by
(10.3.8) and C = 4µb Cnβ + CYβ Cnr . A = 8µ2b KZ2 , B = −2µb Cnr + 2KZ2 CYβ And that concludes our discussion on the modes of vibration. 50 10.3.5 Stability criteria From the characteristic equation (equarion (10.1.4)) we can see which eigenmotions are stable. We have seen earlier that, if A, B, C, D, E and R are all positive, then all eigenmotions are stable. In other words, we have spiral stability and a convergent Dutch roll. However, if some of the coefficients become negative, then there will be unstable eigenmotions. If E < 0, then we have spiral instability. Similarly, if R < 0, then we will have a divergent Dutch roll. So, to ensure stability, we’d best keep the coefficient of the characteristic equation positive. 51