12.5 Trim, Linearization & Longitudinal Modes split the linearised aircraft into two blocks and studied the one in the plane of symmetry. This lesson takes the other: everything that tips the wings or swings the nose sideways. The method is identical (nudge each state in the simulator’s derivative function, collect a matrix, read its eigenvalues), and out come three modes: a roll that stops as soon as you let go of the stick, a slow spiral that may quietly tighten into a dive, and the Dutch roll, a nose-wag-and-wing-rock that drove engineers to build the first yaw dampers.
Tools: 12.4 6-DOF Aircraft Equations of Motion, 7.4 Complex & Repeated Eigenvalues, 1.7 Complex Eigenvalues & Rotation.
Transcript
- Last lesson's pitch block had two clocks. This block has three. After an aileron pulse the roll rate settles in a fraction of a second. A sideslip kick makes the nose wag and fade in seconds. A released bank keeps turning for minutes. Four eigenvalues give all three, before any time stepping.
- At wings-level trim the Jacobian splits in two, because of symmetry. A slip left or right changes lift, drag and pitching moment equally, so their first derivative is zero. What remains is side velocity v, roll rate p, yaw rate r and bank phi. Heading and position drop out.
- At the 12.5 trim the lateral block is a four by four matrix. Row 2, column 1 is minus 0.3175: the dihedral effect. Row 2, column 2, minus 9.428 per second, is roll damping. Row 3, column 1 is 0.0905, the weathercock stability of the fin, and row 3, column 3, minus 0.782, is yaw damping.
- The characteristic polynomial has three families of roots. Minus 9.3790 is a pure exponential with time constant 0.1066 seconds: roll subsidence. Minus 0.4832 plus or minus 2.2898 i is the Dutch roll, zeta 0.2065, period 2.74 seconds. And minus 0.00678 sits almost on the axis: the spiral, 147.5 seconds.
- Roll subsidence first. Hold the stick right and roll damping grows with the roll rate until it cancels the aileron, so the rate follows one minus e to the minus t over tau. Five degrees of aileron gives a steady 17.21 degrees per second; the full model peaks at 16.74. Centre the stick and the rate stops, but the bank stays: 14.48 degrees after ten seconds.
- Why does the bank stay, and which way does it drift? Bank right, and the tilted lift pulls the aircraft into a slip to the right. Dihedral rolls it back toward level. But the fin swings the nose into the slip, and a yawing aircraft rolls further right. If the fin wins, the bank tightens: the spiral divergence.
- Near zero only D lambda plus E matters, so the spiral root is about minus E over D: minus 0.006765, against an exact minus 0.006778. E has two parts. Dihedral against the fin gives 0.7775, a contest of 0.2483 levelling and 0.1690 steepening. Nose-up flight takes off 0.4293, leaving E equal 0.3482: stable, by a small margin.
- Designers cannot just add dihedral. Make it minus 0.2 and the spiral becomes firmly stable, minus 0.0664, but the Dutch roll zeta falls from 0.2065 to 0.184. Enlarge the fin to 0.12 and the Dutch roll stiffens, period 2.04 seconds, but the spiral root turns positive, 0.0177: it diverges.
- The Dutch roll. The fin is a spring pulling the nose back into the wind; yaw inertia carries it past, and the nose wags. Each slip rolls the wings through dihedral. The only strong dashpot is the fin's sideways swing, so damping is weak: after a sideslip kick the peaks 1.32, 0.35 and 0.089 degrees are 2.74 seconds apart, and each cycle divides the amplitude by 3.77.
- Keep only slip and yaw. Omega n squared is Y v times N r plus u zero times N v, and the fin term, 4.52, dominates. By hand, omega n is 2.153 and zeta 0.215, within 8 percent of the full matrix's 2.3402 and 0.2065.
- Now the yaw damper. A rate gyro measures yaw rate and the rudder moves against it: delta r equals minus K r. That is artificial yaw damping, N r minus 5.0643 K. Sweep K: at 0.2, zeta is 0.410; at 0.5, 0.733. In a rudder pulse at 0.5 the peak sideslip falls from 6.27 to 3.64 degrees, with at most 2.49 degrees of damper rudder.
- But a plain damper fights turns. In a steady turn the yaw rate is not zero, so it holds opposite rudder, and the spiral root jumps to minus 0.173. Release a ten degree bank: open loop turns 145 degrees in 120 seconds; with K 0.5 only 11. A washout filter passes the wobble but not the turn: zeta 0.525, spiral untouched, 132 degrees. That is the classic yaw damper.
- To recap. The lateral block is a four by four matrix with three modes. Roll subsidence, minus 9.379, dies in a tenth of a second. The Dutch roll, zeta 0.2065, is the fin as a spring. The spiral, dihedral against the fin, takes 147.5 seconds. A yaw damper lifts zeta to 0.733, and a washout keeps it from fighting turns.
- Next, orbital mechanics. Leave the atmosphere: no lift, no fin, no derivatives. The same F equals m a, with gravity's inverse square law, and the conserved quantities that make orbits close.
Why the lateral block stands alone
At wings-level trim the Jacobian splits in two, and the reason is symmetry. A small slip to the left and one to the right change lift, drag and pitching moment equally, so their first derivative is zero: the longitudinal equations cannot feel , , or to first order. Conversely, a change in speed or angle of attack produces no side force, rolling or yawing moment on a symmetric aircraft flying straight.
The lateral states are side velocity (sideslip ), roll rate , yaw rate and bank . Heading and sideways position drop out: nothing in the physics cares which way north is.
The lateral matrix
The sim builds 12.4’s 13-state vector at the 12.5 trim ( m/s, ), with the quaternion for Euler angles , perturbs one lateral state at a time by central differences, and keeps , , . The last row is 12.3’s kinematics, . Columns :
The same differences on the controls give the aileron column and the rudder column , per radian.
Reading the entries. The derivatives are 12.4’s stated assumptions; nothing new is added.
- , : the rotating-frame terms . Yawing right swings the nose across the airflow, building side velocity at .
- : bank tilts the lift, and the weight pulls sideways. This entry connects bank to sideslip.
- : side force against a slip.
- : dihedral effect, . A slip to the right raises the angle of attack of the right, upwind wing (tilted up toward the wind) and rolls the aircraft left, away from the slip.
- s: roll damping, . The wing going down meets air coming up at it, gains angle of attack and lift, and resists.
- : weathercock stability, . The fin is a weathervane: a slip to the right pushes the tail left and swings the nose right, into the relative wind.
- s: yaw damping, , mostly the fin swinging sideways through the air.
Three eigenvalues, three motions
The characteristic polynomial is . Its roots (sim readout, matched by the lab cell):
| mode | (1/s) | time constant | (rad/s) | period | |
|---|---|---|---|---|---|
| roll subsidence | s | ||||
| Dutch roll | s | ||||
| spiral | s |
A real root is a pure exponential, with time constant ; the pair turns at rad/s and halves every s.
Mode shapes (eigenvectors per radian of bank, sim readout): the roll is almost pure roll rate (, ). The spiral has and rad/s, close to the of a coordinated turn: a slowly changing turn with almost no slip. The Dutch roll has : the wings rock rad per radian of slip.
Roll subsidence
Hold the stick right: the aileron accelerates the roll, and roll damping grows with until it cancels the aileron. With only :
Aileron sets the roll rate, not the bank.
Worked example: the roll time constant by hand
- with N: s.
- s; the gives s.
- For : rad/s /s.
- After one second, .
The sim’s aileron pulse ( for s) peaks at /s at s and reaches at s: the one-state model is high, because the roll also slips the aircraft and dihedral pushes back. Then the stick is centred and the roll stops. The bank stays: at s it is , the heading has swung , and with a spiral root of nothing levels the wings on a human time scale.
The spiral: dihedral against the fin
Why does the bank stay, and which way does it drift? Bank right and the tilted lift pulls the aircraft sideways to the right: a slip to the right, . Two things answer the slip:
- Dihedral () rolls the aircraft left, back toward wings level.
- The fin () swings the nose right, into the slip. That is a yaw rate , and yawing right makes the left wing move faster than the right, so rolls the aircraft further right. Yaw damping slows the swing.
A strong fin turns the slip into yaw, and the yaw into more bank: a tightening, descending turn, the spiral divergence. Strong dihedral levels the wings first.
The spiral root from the polynomial
The spiral root is tiny, so near it , and are negligible beside :
against the exact . The sign of decides the spiral (here ). For a , ; expanding down the last column, which has only in it:
The first bracket is the contest above: dihedral times yaw damping (, levelling) against weathercock times the yaw-roll coupling (, steepening). The often-quoted test is this bracket alone. The second term comes from flying nose-up in body axes: a steady level turn about the vertical appears, about the tilted body axis, as a small roll rate the other way, and roll damping resists that by rolling further into the turn. Here it costs more than half the margin.
is linear in (Problem 2), and the spiral is neutral at , where the simple test would say ; on the fin side at . This aircraft is spirally stable by a small margin.
The Dutch roll
Kick the nose right with the rudder. The fin is a spring pulling the nose back into the wind; yaw inertia carries it past; the nose wags. Each wag is a slip, and dihedral turns each slip into a roll, so the wings rock with the nose, like a skater swinging from edge to edge (the usual story for the name, from memory). The only strong dashpot is the fin’s sideways swing, so the damping is weak.
Keeping only slip and yaw, the block gives , the fin term s dominating, and : rad/s, , slower than the (Problem 1).
Decay per cycle. One period multiplies the amplitude by . The sim’s sideslip kick () has peaks of , and at , and s: spacing s and ratios and . The nose ends right, swung into the wind like a weathervane. The rudder pulse gives peaks and s apart and leaves the aircraft in a bank: yaw became slip, slip became roll.
The yaw damper
The Dutch roll lacks a dashpot in yaw, so build one: measure with a rate gyro and move the rudder against it, . With in seconds (degrees of rudder per degree per second of yaw), the loop changes one column of :
That is artificial yaw damping. The sim’s slider moves the closed-loop roots:
| (s) | Dutch roll | spiral | |
|---|---|---|---|
| , washout |
The pair moves left and down; past it reaches the real axis (the sim reports the roots merged). In the rudder-pulse run at the largest sideslip falls from to , using at most of damper rudder, and one small overshoot ( at s) ends the wag.
But it fights turns. In a steady turn is not zero, so a plain damper holds opposite rudder all the way round. The spiral root jumps from to : the damper rolls the aircraft out of every turn. Release a bank and in s the open-loop aircraft turns ; with the wings are level and the heading has changed only .
The washout filter. Feed the damper , with , s. is a lagged copy of , so passes the s wobble but decays to zero in a steady turn. It adds a fifth root ( at ). The Dutch roll gets instead of , but the spiral stays at and the bank release turns . This is the classic yaw damper.
How small is small?
The sim flies the nonlinear model beside the linear one. After the rudder pulse the largest gaps in 10 s are of sideslip and of bank. After the aileron pulse the bank gap reaches of , and the nonlinear aircraft has lost m: banked lift no longer balances the weight.
The spiral shows the limit sharply. Set (spiral root , doubling every s) and release a bank. At s the nonlinear bank is , the linear one , and m of height is gone; at s the nonlinear aircraft is at of bank, m/s and m lower. The linear model detects the instability but cannot follow it.
Handling qualities
From memory, MIL-F-8785C asks small aircraft for Level 1 Dutch roll , rad/s, rad/s; a roll time constant under about to s; and allows an unstable spiral that takes at least to s to double, depending on the task. Ours (, s, stable spiral) pass. Even the weak-dihedral variant, doubling in s, would pass: pilots correct a slow divergence without noticing.
Computing it
This is 12.4’s model with , (the gyroscopic terms then drop out of roll and yaw), so it reproduces the sim’s matrix. Try Clb = -0.03: the spiral root turns positive.
Graded: the three modes, and a weak-dihedral matrix (; only the column changes) whose spiral root is positive.
Concept checks
Problems
Keep only slip and yaw, , , with , , , . (a) Find , and the period, and compare with the . (b) By what factor does the amplitude fall per cycle, using the root ? (c) Compare with the sim’s slip-kick peaks , , .
Show worked solution
(a) The characteristic polynomial of the is , so , rad/s, and , ; , rad/s, period s. The gives rad/s, , s: dropping roll and bank costs in frequency.
(b) s, factor .
(c) and : close, the small excess from the spiral and roll content of the start.
(a) Expand down its last column to show . (b) Explain why the smallest root is about . (c) The sim gives at and at . Why is linear in , and where is the spiral neutral?
Show worked solution
(a) Column 4 is , so , with the determinant of rows –, columns –: . Multiply out: .
(b) For tiny , is negligible beside , so : against the exact .
(c) enters only through , so it changes only and (via ), both linearly, and has no product of two such terms. Interpolating: .
With the rudder column changes column of . (a) Write the block for . (b) Predict the Dutch-roll . (c) The sim gives ; why is the prediction high?
Show worked solution
(a) , and (the rudder’s side force).
(b) , ; .
(c) As without the damper ( against ), the drops roll and bank. The gap grows from to as the pair nears the spiral root (now ); at they merge.
Key Takeaways
- Symmetry at wings-level trim makes a system of its own, found by 12.5’s central-difference Jacobian.
- Three modes: roll subsidence ( s), Dutch roll (, s), spiral ( s).
- Roll damping makes the aileron command a roll rate; the bank then stays.
- The spiral is dihedral against the fin; its root is about with , neutral here at or .
- The Dutch roll is the fin as a spring, coupled into roll by dihedral.
- A yaw damper adds yaw damping ( at ) but fights turns; a washout filter keeps the damping () and leaves the spiral alone.
- Next, 12.7 Orbital Mechanics: leave the atmosphere. No lift, no fin, no derivatives: the same with a different force law, gravity’s inverse square, and the conserved quantities that make orbits close.