Stability is the ability of the aircraft to counteract disturbances to its flight path. According to
David P. Davies, there are six types of aircraft stability: speed stability, stick free static longitudinal stability, static lateral stability, directional stability, oscillatory stability, and spiral stability.
Speed stability An aircraft in
cruise flight is typically speed stable. If speed increases, drag increases, which will reduce the speed back to equilibrium for its configuration and thrust setting. If speed decreases, drag decreases, and the aircraft will accelerate back to its equilibrium speed where thrust equals drag. However, in slow flight, due to
lift-induced drag, as speed decreases, drag increases (and vice versa). This is known as the "back of the
drag curve". The aircraft will be speed unstable, because a decrease in speed will cause a further decrease in speed.
Static stability and control Longitudinal static stability Longitudinal stability refers to the stability of an aircraft in pitch. For a stable aircraft, if the aircraft pitches up, the wings and tail create a pitch-down moment which tends to restore the aircraft to its original attitude. For an unstable aircraft, a disturbance in pitch will lead to an increasing pitching moment. Longitudinal static stability is the ability of an aircraft to recover from an initial disturbance. Longitudinal dynamic stability refers to the damping of these stabilizing moments, which prevents persistent or increasing oscillations in pitch.
Directional stability Directional or weathercock stability is concerned with the
static stability of the airplane about the z axis. Just as in the case of longitudinal stability it is desirable that the aircraft should tend to return to an equilibrium condition when subjected to some form of yawing disturbance. For this the slope of the yawing moment curve must be positive. An airplane possessing this mode of stability will always point towards the
relative wind, hence the name weathercock stability.
Dynamic stability and control Longitudinal modes It is common practice to derive a fourth order
characteristic equation to describe the longitudinal motion, and then factorize it approximately into a high frequency mode and a low frequency mode. The approach adopted here is using qualitative knowledge of aircraft behavior to simplify the equations from the outset, reaching the result by a more accessible route. The two longitudinal motions (modes) are called the
short period pitch oscillation (SPPO), and the
phugoid.
Short-period pitch oscillation A short input (in
control systems terminology an
impulse) in pitch (generally via the elevator in a standard configuration fixed-wing aircraft) will generally lead to overshoots about the trimmed condition. The transition is characterized by a
damped simple harmonic motion about the new trim. There is very little change in the trajectory over the time it takes for the oscillation to damp out. Generally this oscillation is high frequency (hence short period) and is damped over a period of a few seconds. A real-world example would involve a pilot selecting a new climb attitude, for example 5° nose up from the original attitude. A short, sharp pull back on the control column may be used, and will generally lead to oscillations about the new trim condition. If the oscillations are poorly damped the aircraft will take a long period of time to settle at the new condition, potentially leading to
Pilot-induced oscillation. If the short period mode is unstable it will generally be impossible for the pilot to safely control the aircraft for any period of time. This damped harmonic motion is called the
short period pitch oscillation; it arises from the tendency of a stable aircraft to point in the general direction of flight. It is very similar in nature to the
weathercock mode of missile or rocket configurations. The motion involves mainly the pitch attitude \theta (theta) and incidence \alpha (alpha). The direction of the velocity vector, relative to inertial axes is \theta-\alpha. The velocity vector is: ::u_f=U\cos(\theta-\alpha) ::w_f=U\sin(\theta-\alpha) where u_f, w_f are the inertial axes components of velocity. According to
Newton's second law, the
accelerations are proportional to the
forces, so the forces in inertial axes are: ::X_f=m\frac{du_f}{dt}=m\frac{dU}{dt}\cos(\theta-\alpha)-mU\frac{d(\theta-\alpha)}{dt}\sin(\theta-\alpha) ::Z_f=m\frac{dw_f}{dt}=m\frac{dU}{dt}\sin(\theta-\alpha)+mU\frac{d(\theta-\alpha)}{dt}\cos(\theta-\alpha) where
m is the
mass. By the nature of the motion, the speed variation m\frac{dU}{dt} is negligible over the period of the oscillation, so: ::X_f= -mU\frac{d(\theta-\alpha)}{dt}\sin(\theta-\alpha) ::Z_f=mU\frac{d(\theta-\alpha)}{dt}\cos(\theta-\alpha) But the forces are generated by the
pressure distribution on the body, and are referred to the velocity vector. But the velocity (wind) axes set is not an
inertial frame so we must resolve the fixed axes forces into wind axes. Also, we are only concerned with the force along the z-axis: ::Z=-Z_f\cos(\theta-\alpha)+X_f\sin(\theta-\alpha) Or: ::Z=-mU\frac{d(\theta-\alpha)}{dt} In words, the wind axes force is equal to the
centripetal acceleration. The moment equation is the
time derivative of the
angular momentum: ::M=B\frac{d^2 \theta}{dt^2} where M is the pitching moment, and B is the
moment of inertia about the pitch axis. Let: \frac{d\theta}{dt}=q, the pitch rate. The equations of motion, with all forces and moments referred to wind axes are, therefore: ::\frac{d\alpha}{dt}=q+\frac{Z}{mU} ::\frac{dq}{dt}=\frac{M}{B} We are only concerned with perturbations in forces and moments, due to perturbations in the states \alpha and q, and their time derivatives. These are characterized by
stability derivatives determined from the flight condition. The possible stability derivatives are: ::Z_\alpha Lift due to incidence, this is negative because the z-axis is downwards whilst positive incidence causes an upwards force. ::Z_q Lift due to pitch rate, arises from the increase in tail incidence, hence is also negative, but small compared with Z_\alpha. ::M_\alpha
Pitching moment due to incidence - the static stability term.
Static stability requires this to be negative. ::M_q Pitching moment due to pitch rate - the pitch damping term, this is always negative. Since the tail is operating in the flowfield of the wing, changes in the wing incidence cause changes in the downwash, but there is a delay for the change in wing flowfield to affect the tail lift, this is represented as a moment proportional to the rate of change of incidence: ::M_\dot\alpha The delayed downwash effect gives the tail more lift and produces a nose down moment, so M_\dot\alpha is expected to be negative. The equations of motion, with small perturbation forces and moments become: ::\frac{d\alpha}{dt}=\left(1+\frac{Z_q}{mU}\right)q+\frac{Z_\alpha}{mU}\alpha ::\frac{dq}{dt}=\frac{M_q}{B}q+\frac{M_\alpha}{B}\alpha+\frac{M_\dot\alpha}{B}\dot\alpha These may be manipulated to yield as second order linear
differential equation in \alpha: ::\frac{d^2\alpha}{dt^2}-\left(\frac{Z_\alpha}{mU}+\frac{M_q}{B}+(1+\frac{Z_q}{mU})\frac{M_\dot\alpha}{B}\right)\frac{d\alpha}{dt}+\left(\frac{Z_\alpha}{mU}\frac{M_q}{B}-\frac{M_\alpha}{B}(1+\frac{Z_q}{mU})\right)\alpha=0 This represents a damped simple harmonic motion. We should expect \frac{Z_q}{mU} to be small compared with unity, so the coefficient of \alpha (the 'stiffness' term) will be positive, provided M_\alpha. This expression is dominated by M_\alpha, which defines the
longitudinal static stability of the aircraft, it must be negative for stability. The damping term is reduced by the downwash effect, and it is difficult to design an aircraft with both rapid natural response and heavy damping. Usually, the response is underdamped but stable.
Phugoid If the stick is held fixed, the aircraft will not maintain straight and level flight (except in the unlikely case that it happens to be perfectly trimmed for level flight at its current altitude and thrust setting), but will start to dive, level out and climb again. It will repeat this cycle until the pilot intervenes. This long period oscillation in speed and height is called the
phugoid mode. This is analyzed by assuming that the
SSPO performs its proper function and maintains the angle of attack near its nominal value. The two states which are mainly affected are the flight path angle \gamma (gamma) and speed. The small perturbation equations of motion are: ::mU\frac{d\gamma}{dt}=-Z which means the centripetal force is equal to the perturbation in lift force. For the speed, resolving along the trajectory: ::m\frac{du}{dt}=X-mg\gamma where g is the
acceleration due to gravity at the Earth's surface. The acceleration along the trajectory is equal to the net x-wise force minus the component of weight. We should not expect significant aerodynamic derivatives to depend on the flight path angle, so only X_u and Z_u need be considered. X_u is the drag increment with increased speed, it is negative, likewise Z_u is the lift increment due to speed increment, it is also negative because lift acts in the opposite sense to the z-axis. The equations of motion become: :: mU\frac{d\gamma}{dt}=-Z_u u :: m\frac{du}{dt}=X_u u -mg\gamma These may be expressed as a second order equation in flight path angle or speed perturbation: ::\frac{d^2u}{dt^2}-\frac{X_u}{m}\frac{du}{dt}-\frac{Z_ug}{mU}u=0 Now lift is very nearly equal to weight: ::Z=\frac{1}{2}\rho U^2 c_L S_w=W where \rho is the air density, S_w is the wing area, W the weight and c_L is the lift coefficient (assumed constant because the incidence is constant), we have, approximately: ::Z_u=\frac{2W}{U}=\frac{2mg}{U} The period of the phugoid, T, is obtained from the coefficient of u: ::\frac{2\pi}{T}=\sqrt{\frac{2g^2}{U^2}} Or: ::T=\frac{2\pi U}{\sqrt{2}g} Since the lift is very much greater than the drag, the phugoid is at best lightly damped. A
propeller with fixed speed would help. Heavy damping of the pitch rotation or a large
rotational inertia increase the coupling between short period and phugoid modes, so that these will modify the phugoid.
Lateral modes had
anhedral wings, which make it less stable but more manoeuvrable. With a symmetrical rocket or missile, the
directional stability in yaw is the same as the pitch stability; it resembles the short period pitch oscillation, with yaw plane equivalents to the pitch plane stability derivatives. For this reason, pitch and yaw directional stability are collectively known as the "weathercock" stability of the missile. Aircraft lack the symmetry between pitch and yaw, so that directional stability in yaw is derived from a different set of stability derivatives. The yaw plane equivalent to the short period pitch oscillation, which describes yaw plane directional stability is called Dutch roll. Unlike pitch plane motions, the lateral modes involve both roll and yaw motion.
Dutch roll It is customary to derive the equations of motion by formal manipulation in what, to the engineer, amounts to a piece of mathematical sleight of hand. The current approach follows the pitch plane analysis in formulating the equations in terms of concepts which are reasonably familiar. Applying an impulse via the rudder pedals should induce
Dutch roll, which is the oscillation in roll and yaw, with the roll motion lagging yaw by a quarter cycle, so that the wing tips follow elliptical paths with respect to the aircraft. The yaw plane translational equation, as in the pitch plane, equates the centripetal acceleration to the side force. ::\frac{d\beta}{dt}=\frac{Y}{mU}-r where \beta (beta) is the
sideslip angle, Y the side force and r the yaw rate. The moment equations are a bit trickier. The trim condition is with the aircraft at an angle of attack with respect to the airflow. The body x-axis does not align with the velocity vector, which is the reference direction for wind axes. In other words, wind axes are not
principal axes (the mass is not distributed symmetrically about the yaw and roll axes). Consider the motion of an element of mass in position -z, x in the direction of the y-axis, i.e. into the plane of the paper. If the roll rate is p, the velocity of the particle is: :::v=-pz+xr Made up of two terms, the force on this particle is first the proportional to rate of v change, the second is due to the change in direction of this component of velocity as the body moves. The latter terms gives rise to cross products of small quantities (pq, pr, qr), which are later discarded. In this analysis, they are discarded from the outset for the sake of clarity. In effect, we assume that the direction of the velocity of the particle due to the simultaneous roll and yaw rates does not change significantly throughout the motion. With this simplifying assumption, the acceleration of the particle becomes: :::\frac{dv}{dt}=-\frac{dp}{dt}z+\frac{dr}{dt}x The yawing moment is given by: :::\delta m x \frac{dv}{dt}=-\frac{dp}{dt}xz\delta m + \frac{dr}{dt}x^2\delta m There is an additional yawing moment due to the offset of the particle in the y direction:\frac{dr}{dt}y^2\delta m The yawing moment is found by summing over all particles of the body: :::N=-\frac{dp}{dt}\int xz dm +\frac{dr}{dt}\int x^2 + y^2 dm =-E\frac{dp}{dt}+C\frac{dr}{dt} where N is the yawing moment, E is a product of inertia, and C is the moment of inertia about the
yaw axis. A similar reasoning yields the roll equation: :::L=A\frac{dp}{dt}-E\frac{dr}{dt} where L is the rolling moment and A the roll moment of inertia.
Lateral and longitudinal stability derivatives The states are \beta (sideslip), r (yaw rate) and p (roll rate), with moments N (yaw) and L (roll), and force Y (sideways). There are nine stability derivatives relevant to this motion, the following explains how they originate. However a better intuitive understanding is to be gained by simply playing with a model airplane, and considering how the forces on each component are affected by changes in sideslip and
angular velocity: :::Y_\beta Side force due to side slip (in absence of yaw). Sideslip generates a sideforce from the fin and the fuselage. In addition, if the wing has dihedral, side slip at a positive roll angle increases incidence on the starboard wing and reduces it on the port side, resulting in a net force component directly opposite to the sideslip direction. Sweep back of the wings has the same effect on incidence, but since the wings are not inclined in the vertical plane, backsweep alone does not affect Y_\beta. However, anhedral may be used with high backsweep angles in high performance aircraft to offset the wing incidence effects of sideslip. Oddly enough this does not reverse the sign of the wing configuration's contribution to Y_\beta (compared to the dihedral case). :::Y_p Side force due to roll rate. Roll rate causes incidence at the fin, which generates a corresponding side force. Also, positive roll (starboard wing down) increases the lift on the starboard wing and reduces it on the port. If the wing has dihedral, this will result in a side force momentarily opposing the resultant sideslip tendency. Anhedral wing and or stabilizer configurations can cause the sign of the side force to invert if the fin effect is swamped. :::Y_r Side force due to yaw rate. Yawing generates side forces due to incidence at the rudder, fin and fuselage. :::N_\beta Yawing moment due to sideslip forces. Sideslip in the absence of rudder input causes incidence on the fuselage and
empennage, thus creating a yawing moment counteracted only by the directional stiffness which would tend to point the aircraft's nose back into the wind in horizontal flight conditions. Under sideslip conditions at a given roll angle N_\beta will tend to point the nose into the sideslip direction even without rudder input, causing a downward spiraling flight. :::N_p Yawing moment due to roll rate. Roll rate generates fin lift causing a yawing moment and also differentially alters the lift on the wings, thus affecting the induced drag contribution of each wing, causing a (small) yawing moment contribution. Positive roll generally causes positive N_p values unless the
empennage is anhedral or fin is below the roll axis. Lateral force components resulting from dihedral or anhedral wing lift differences has little effect on N_p because the wing axis is normally closely aligned with the center of gravity. :::N_r Yawing moment due to yaw rate. Yaw rate input at any roll angle generates rudder, fin and fuselage force vectors which dominate the resultant yawing moment. Yawing also increases the speed of the outboard wing whilst slowing down the inboard wing, with corresponding changes in drag causing a (small) opposing yaw moment. N_r opposes the inherent directional stiffness which tends to point the aircraft's nose back into the wind and always matches the sign of the yaw rate input. :::L_\beta Rolling moment due to sideslip. A positive sideslip angle generates empennage incidence which can cause positive or negative roll moment depending on its configuration. For any non-zero sideslip angle dihedral wings causes a rolling moment which tends to return the aircraft to the horizontal, as does back swept wings. With highly swept wings the resultant rolling moment may be excessive for all stability requirements and anhedral could be used to offset the effect of wing sweep induced rolling moment. :::L_r Rolling moment due to yaw rate. Yaw increases the speed of the outboard wing whilst reducing speed of the inboard one, causing a rolling moment to the inboard side. The contribution of the fin normally supports this inward rolling effect unless offset by anhedral stabilizer above the roll axis (or dihedral below the roll axis). :::L_p Rolling moment due to roll rate. Roll creates counter rotational forces on both starboard and port wings whilst also generating such forces at the empennage. These opposing rolling moment effects have to be overcome by the aileron input in order to sustain the roll rate. If the roll is stopped at a non-zero roll angle the L_\beta
upward rolling moment induced by the ensuing sideslip should return the aircraft to the horizontal unless exceeded in turn by the
downward L_r rolling moment resulting from sideslip induced yaw rate. Longitudinal stability could be ensured or improved by minimizing the latter effect.
Equations of motion Since
Dutch roll is a handling mode, analogous to the short period pitch oscillation, any effect it might have on the trajectory may be ignored. The body rate
r is made up of the rate of change of sideslip angle and the rate of turn. Taking the latter as zero, assuming no effect on the trajectory, for the limited purpose of studying the Dutch roll: :::\frac{d\beta}{dt}= -r The yaw and roll equations, with the stability derivatives become: ::C\frac{dr}{dt}-E\frac{dp}{dt}=N_\beta \beta - N_r \frac{d\beta}{dt} + N_p p (yaw) ::A\frac{dp}{dt}-E\frac{dr}{dt}=L_\beta \beta - L_r \frac{d\beta}{dt} + L_p p (roll) The inertial moment due to the roll acceleration is considered small compared with the aerodynamic terms, so the equations become: ::-C\frac{d^2\beta}{dt^2} = N_\beta \beta - N_r \frac{d\beta}{dt} + N_p p ::E\frac{d^2\beta}{dt^2} = L_\beta \beta - L_r \frac{d\beta}{dt} + L_p p This becomes a second order equation governing either roll rate or sideslip: ::\left(\frac{N_p}{C}\frac{E}{A}-\frac{L_p}{A}\right)\frac{d^2\beta}{dt^2}+ \left(\frac{L_p}{A}\frac{N_r}{C}-\frac{N_p}{C}\frac{L_r}{A}\right)\frac{d\beta}{dt}- \left(\frac{L_p}{A}\frac{N_\beta}{C}-\frac{L_\beta}{A}\frac{N_p}{C}\right)\beta = 0 The equation for roll rate is identical. But the roll angle,
\phi (phi) is given by: :::\frac{d\phi}{dt}=p If
p is a damped simple harmonic motion, so is
\phi, but the roll must be in
quadrature with the roll rate, and hence also with the sideslip. The motion consists of oscillations in roll and yaw, with the roll motion lagging 90 degrees behind the yaw. The wing tips trace out elliptical paths. Stability requires the "
stiffness" and "damping" terms to be positive. These are: :::\frac{\frac{L_p}{A}\frac{N_r}{C}-\frac{N_p}{C}\frac{L_r}{A}} {\frac{N_p}{C}\frac{E}{A}-\frac{L_p}{A}} (damping) :::\frac{\frac{L_\beta}{A}\frac{N_p}{C}-\frac{L_p}{A}\frac{N_\beta}{C}} {\frac{N_p}{C}\frac{E}{A}-\frac{L_p}{A}} (stiffness) The denominator is dominated by L_p, the roll damping derivative, which is always negative, so the denominators of these two expressions will be positive. Considering the "stiffness" term: -L_p N_\beta will be positive because L_p is always negative and N_\beta is positive by design. L_\beta is usually negative, whilst N_p is positive. Excessive dihedral can destabilize the Dutch roll, so configurations with highly swept wings require anhedral to offset the wing sweep contribution to L_\beta. The damping term is dominated by the product of the roll damping and the yaw damping derivatives, these are both negative, so their product is positive. The Dutch roll should therefore be damped. The motion is accompanied by slight lateral motion of the center of gravity and a more "exact" analysis will introduce terms in Y_\beta etc. In view of the accuracy with which stability derivatives can be calculated, this is an unnecessary pedantry, which serves to obscure the relationship between aircraft geometry and handling, which is the fundamental objective of this article.
Roll subsidence Jerking the stick sideways and returning it to center causes a net change in roll orientation. The roll motion is characterized by an absence of natural stability, there are no stability derivatives which generate moments in response to the inertial roll angle. A roll disturbance induces a roll rate which is only canceled by pilot or
autopilot intervention. This takes place with insignificant changes in sideslip or yaw rate, so the equation of motion reduces to: ::A\frac{dp}{dt}=L_p p. L_p is negative, so the roll rate will decay with time. The roll rate reduces to zero, but there is no direct control over the roll angle.
Spiral mode Simply holding the stick still, when starting with the wings near level, an aircraft will usually have a tendency to gradually veer off to one side of the straight flightpath. This is the (slightly unstable)
spiral mode.
Spiral mode trajectory In studying the trajectory, it is the direction of the velocity vector, rather than that of the body, which is of interest. The direction of the velocity vector when projected on to the horizontal will be called the track, denoted
\mu (
mu). The body orientation is called the heading, denoted
\psi (psi). The force equation of motion includes a component of weight: ::\frac{d\mu}{dt}=\frac{Y}{mU} + \frac{g}{U}\phi where
g is the gravitational acceleration, and
U is the speed. Including the stability derivatives: ::\frac{d\mu}{dt}=\frac{Y_\beta}{mU}\beta + \frac {Y_r}{mU}r + \frac{Y_p}{mU}p + \frac{g}{U}\phi Roll rates and yaw rates are expected to be small, so the contributions of Y_r and Y_p will be ignored. The sideslip and roll rate vary gradually, so their time
derivatives are ignored. The yaw and roll equations reduce to: ::N_\beta \beta + N_r\frac{d\mu}{dt} + N_p p = 0 (yaw) ::L_\beta \beta + L_r\frac{d\mu}{dt} + L_p p = 0 (roll) Solving for
\beta and
p: :::\beta=\frac{(L_r N_p - L_p N_r)}{(L_p N_\beta - N_p L_\beta)}\frac{d\mu}{dt} :::p=\frac{(L_\beta N_r - L_r N_\beta)}{(L_p N_\beta - N_p L_\beta)}\frac{d\mu}{dt} Substituting for sideslip and roll rate in the force equation results in a first order equation in roll angle: :::\frac{d\phi}{dt}=mg\frac{(L_\beta N_r - N_\beta L_r)}{mU(L_p N_\beta - N_p L_\beta)-Y_\beta(L_r N_p - L_p N_r)}\phi This is an
exponential growth or decay, depending on whether the coefficient of
\phi is positive or negative. The denominator is usually negative, which requires L_\beta N_r > N_\beta L_r (both products are positive). This is in direct conflict with the Dutch roll stability requirement, and it is difficult to design an aircraft for which both the Dutch roll and spiral mode are inherently stable. Since the
spiral mode has a long time constant, the pilot can intervene to effectively stabilize it, but an aircraft with an unstable Dutch roll would be difficult to fly. It is usual to design the aircraft with a stable Dutch roll mode, but slightly unstable spiral mode. == See also ==