Roll center migration: the number that won't sit still

Your roll center is a design value at exactly one ride height. Drive the car and it moves — on one Mini, by 186 mm across travel, straight through ground level. Here's how to measure the envelope and how much is too much.

Last chapter ended on a warning: the roll-center construction is valid at one position. This chapter is what happens when you ignore that.

Here is the strut layout’s roll center, measured straight out of the solver at each wheel position:

Wheel travel	Roll center
−40 mm (droop)	158 mm
0 mm (static)	90 mm
+20 mm	54 mm
+40 mm	14 mm
+60 mm	−28 mm

The number you’d put on a spec sheet is 90 mm. The number the car actually uses ranges from 158 mm to below ground — an envelope of 186 mm, which is larger than the value itself.

Now recall the roll-moment equation from chapter 4: roll resistance depends on $h_{\text{cg}} - h_{\text{rc}}$ . If the roll center moves 186 mm, that lever changes by 186 mm too. The car’s roll stiffness is not a constant. It is a function of how compressed the suspension happens to be, which is a function of how hard you are already cornering.

That is why a car can feel benign on turn-in and strange at the apex. Nothing failed. The geometry moved.

Why it moves so much

The roll center is derived from the instant center, and you already know from chapter 2 that the IC runs around. Worse, the roll center amplifies it: it’s found by projecting a line from the IC to the contact patch and reading where it crosses the centerline, so a small angular change at a distant IC swings the crossing point a long way.

Rough rule: roll-center movement is the IC’s movement scaled by how far inboard the IC sits. A distant IC gives a stable swing arm and a twitchy roll center. A close IC gives an aggressive camber curve and a steadier roll center. They pull in opposite directions, which is exactly why this is hard.

You can watch that trade directly, on one layout, by moving one point. Take the wishbone layout and shift its lower inner mount up 15 mm, flattening the lower arm:

Lower inner mount	Static FVSA	RC range over travel	Camber gain
as designed	2999 mm	100 mm	0.0196 °/mm
15 mm higher	1855 mm	87 mm	0.0315 °/mm

Shorter swing arm, steadier roll center, and more camber gain — all from one 15 mm move. That is the opposite of the intuition most people arrive with, which says a short swing arm must make everything twitchier.

Push the same point 20 mm the other way and the arms come close to parallel: FVSA balloons to about 18 900 mm, camber gain nearly vanishes at 0.0036 °/mm, and the roll-center range gets worse — 117 mm. The long swing arm did not buy stability. It bought a roll center that sweeps further while the camber curve does nothing at all.

In roll, it also moves sideways

Everything above is heave: both wheels moving together. In a corner they don’t. The outside wheel compresses, the inside extends, and the two sides no longer have mirror-image geometry — so the construction stops being symmetric and the roll center moves laterally as well as vertically, sometimes right out past the wheels.

This is the part that makes purists dislike roll-center analysis altogether, and they have a point: once the roll center has left the centerline, “the body rolls about this point” is no longer a description of anything physical. It stays useful as a bookkeeping device for comparing designs, provided you never forget that is what it is.

The practical consequence is simple: a car whose roll center migrates a lot has a balance that changes with cornering load. Consistency, not the static number, is what you are designing for.

Go look at it

Open the strut layout.

Look at the SETUP ADVISOR on the right. It reports Roll center travel directly — for this car it flags a large figure and warns that balance will shift with load. That advisory is this entire chapter, computed for you.

Hit ▲▼ Bump and watch RC — Roll Center in the metrics panel. It sweeps from around 158 mm down through zero and keeps going.

Now fix it. The advisory suggests making the lower arm flatter. Drag the Lower Bushing (inner end of the lower arm) so the arm sits closer to horizontal at rest, then re-run ▲▼ Bump. The roll-center sweep shrinks.

Check the bill. FVSA — Swing Arm just changed, so your camber gain from chapter 3 is different. Flattening the arms is exactly the move that lengthens the swing arm and kills camber gain.

Then flatten the lower arm: drag the lower inner mount up about 15 mm and re-run the sweep. The band narrows to roughly 87 mm — and FVSA — Swing Arm shortened to about 1855 mm, so you gained camber curve at the same time.

The roll center at five wheel positions, plotted to scale against the ground. The spec sheet quotes the middle one; the car uses the whole band, and on the Mini's front that band is 186 mm tall — larger than the number itself.

Run the sweep yourself — Mini front →

What it costs you

You can always trade migration for camber gain, and the trade is close to linear: arms nearer parallel give a longer swing arm, a steadier roll center, and less camber gain. Arms further from parallel give the reverse.

What you cannot do is escape the trade with a strut. The strut’s perpendicular construction ties the IC’s behavior to the strut angle, and that is the single biggest reason struts have a reputation for vagueness at the limit despite being perfectly good at everything else. It isn’t that the strut is crude. It’s that its roll center is hard to keep still.

Which sets up the honest warning for lowered cars, and the reason chapter 24 exists: lowering a strut car doesn’t merely drop the roll center, it moves the car into the part of its travel where the roll center is moving fastest. Worse static value and worse envelope, in one move.

Rules of thumb

Migration in heave, road car: under 50 mm across usable travel is good; 50–90 mm workable; beyond 90 mm the balance visibly changes with load. The app’s advisory uses similar thresholds.
Circuit car: aim under 30 mm — small travel makes this easier than it sounds.
Rate of change is more useful than the total: under roughly 1 mm of roll-center movement per mm of wheel travel keeps a car feeling consistent. The strut layout above averages about 1.9 mm/mm.
Rally-raid and desert: with 350–600 mm of travel, a small absolute envelope is simply not available — nobody holds a road-car figure over half a meter of stroke. What matters instead is that migration be smooth and monotonic, with no reversal or sudden rate change in the part of the stroke the car actually lives in. A beam axle sidesteps the question: its roll center is set by the Panhard rod or Watt’s linkage, not by control-arm geometry, which is a large part of the appeal at that travel.

Try this

Set the arms as close to parallel as the linkage allows. How small can you make the migration, and what has the camber curve become?
Find the wheel position where the front roll center crosses zero. How much of the car’s normal travel sits below that point?
On the AMG rear, deliberately make the migration worse by tilting the upper arm. How much camber gain do you get per extra 10 mm of roll-center movement?

Next: the tie rod has been sitting quietly in every diagram so far. It’s about to steer your car without being asked.

Preset hardpoints are illustrative sketches, not measured factory specs.