REX-0R Superleggera

Some wheels are designed to a price. This one was designed to a question: how little material can hold a forged monoblock at full load, and where, exactly, does the metal still have to be?

The REX-0R Superleggera is the lightest wheel we forge, and the answer to that question is written into every surface of it. At 18×10.5J ET45 it weighs 7.3 kg per corner — published per corner, in the fitment you actually order, not a best-case figure borrowed from the smallest size in the range. In 17×7J it comes in at 5.6 kg. The weight is not one trick. It is the sum of decisions, each made where the simulation pointed.

Why the long-shaft Y holds its line

Start with what “stability” actually means for a wheel, because it isn’t strength. When you load a car into a corner, the side force at the contact patch tries to tilt the rim away from the hub. If the spokes allow that tilt, the wheel quietly gives away camber and the contact patch wanders — turn-in arrives a beat late, grip won’t sit still. A stable wheel keeps the rim plane true to the hub while the car leans on it. So the thing worth engineering for isn’t how much load the wheel survives — it’s how little it moves: lateral and torsional stiffness, per gram. The REX-0R’s spoke is built around that one number.

Most of the spoke should be the stiff part. A loaded beam deflects according to its length cubed —

δ deflection · L unsupported length · E modulus · I second moment of area

Double the unsupported length and you don’t double the flex, you roughly octuple it. The classic split-spoke wheel forks early, near the hub, and runs two long, slender branches all the way to the rim — so its longest stretch is also its softest, and the L³ term works against it over almost the whole spoke. The REX-0R inverts the layout: one deep main spoke carries the load almost to the rim and splits late, so the long span L is the stiff member and only a short branch length is left soft. Same equation, used the right way round.

That main spoke is a true I-beam. Bending stiffness is the E I in that formula. The material is fixed, so E is fixed — every gain has to come from I, the second moment of area:

b section width · d section depth — stiffness scales with the cube of depth

Push material away from the neutral axis and I climbs with the cube of depth d. We pocket the spoke from each side, scooping the lightly stressed core out of the side faces and leaving the front and back faces as flanges joined by a central web — which sets the section’s depth front-to-back, the axis cornering load bends. On the verified 18in reference build, the REX-0R’s main-spoke mid-span section measures 29.4 mm deep, 25.2 mm wide, with flanges of 4.1 mm front and 5.5 mm back joined by a 13.9 mm web. About that section’s true centroid, bending in the axial plane:

Flanges (47% of the area) carry ~80% of it — back 19,270 · front 17,630 · web only 9,180 mm⁴

Compare the finished section to a solid blank of the same outline. The blank would carry I ≈ 53,400 mm⁴; the pocketed I-beam spoke carries 46,100 mm⁴ (calculated from section geometry) — so milling the core out of the side faces removed 30% of the section’s mass while keeping 86% of its bending stiffness. The metal that left was sitting next to the neutral axis, doing under a seventh of the work; the flanges that stayed carry four-fifths of it. Stiffness per gram rises about 23%. That is the whole design in one measurement: weight gone from where it wasn’t earning its place, rigidity held where the cornering load actually bends the spoke. The same section bends about its tangential axis at only ~17,400 mm⁴ — so the spoke is roughly 2.65× stiffer against lateral load than across it, which is exactly the bias a cornering wheel wants.

The fork is a brace, not a flourish. Where the spoke splits, the two short branches and the rim arc between them close into a wide-based structural triangle. The point is not that a wider included angle reduces branch force for every load; for a load aligned with the spoke, it can actually raise the force each branch carries. The point is triangulation. For the in-plane off-axis loads that decide stability — especially drive and brake torque — a wide-based triangle reacts more of the load through axial tension and compression in its members, and an axially loaded member is far stiffer than a bent one:

axial stiffness AE/L vastly exceeds cantilever bending stiffness 3EI/L³ once L is large compared with the section’s radius of gyration.

A narrow, near-parallel fork has little effective triangulation, so those same in-plane off-axis loads tend to fall back into branch bending — two long cantilevers carrying the load the soft way, under the L³/EI penalty from before. Widen the angle and shorten the branches, and the fork stops behaving like two bending cantilevers and starts behaving more like a braced frame. That is what the short, wide fork buys at the rim; the long, narrow one is a soft cantilever wearing a spoke’s shape.

It’s shaped around the load that usually gets missed. Radial-only simulation never applies the twist, so the wheel looks settled. Add real torque and the spokes deform angularly, with stress piling at the spoke root — the failure path our combined-load method is built to expose. The wide fork answers it directly: two well-separated rim-side branch feet share the torque reaction across a broader base, so the same torque is carried through a wider couple and the force at each landing is correspondingly lower, while the torsionally efficient main spoke ties that reaction back to the hub. The geometry is drawn around the stress path the analysis reveals, not decorated to look like it was.

And it stays composed when inputs come fast. Natural frequency rises with stiffness over modal mass —

modal mass = the effective mass participating in that mode; material near a constrained hub contributes less to many rim/spoke modes because it moves less.

So a stiff structure with its mass kept inboard rather than out at the rim tends to push its relevant resonances upward, away from the low-frequency, high-energy range of transient cornering-and-braking inputs. The inboard placement helps twice over: mass near the hub sits where many mounted-wheel mode shapes have little displacement, so it adds less modal mass and costs less frequency than the same mass placed near the rim. Less flutter, steadier feel. Stiffness you can measure on a bench becomes confidence you feel through the wheel.

The honest trade-off. There’s a reason the classic split-spoke look has lasted: a long, delicate fork is genuinely lovely, and it gives a little — a compliance some drivers like. The REX-0R spends some of that elegance, and some of that give, to buy back lateral and torsional rigidity. On a sports car that’s the right trade every time — the wheel should hold the geometry the chassis was designed around, not add softness of its own. On a show car built to be looked at more than driven, the prettier, more compliant fork may be the better pick, and we’d say so. This wheel is built for the first kind of car.

The same logic runs into the barrel and back to the hub. Where each spoke arches into the barrel, the side beam is deep-pocketed along its length, drawing material off the neutral axis so bending stiffness at the spoke root is held while weight at the join is shed. At the hub, the mounting-pad backface is pocketed around each lug seat — mass removed from behind the bolt circle, where clamp load does not need it, while the seating faces that keep the wheel running true are left completely intact.

At the rim edge sits the Counter-Lip. A second forged bead, set just inboard of the outer flange, forms a stiffening hoop at the part of any wheel most prone to bell-mouth under impact and lateral load. With that hoop in place, the barrel wall inboard of it can run thinner without losing edge rigidity. The displacement study shows why it lives there: under load, the face and spokes stay almost still while deflection concentrates at the outer lip. The Counter-Lip braces the wheel precisely where the wheel wants to move.

None of this is styling, and none of it is taken on faith. The REX-0R was validated under LFI’s combined-load case — radial, torque, lateral and tyre-pressure applied together, not radial-only — at a peak von Mises stress of 100.6 MPa. Our design rule is a one-third-yield target: peak combined-load stress held to roughly a third of the material’s yield, a design factor of three. Against the supplier-rated 313 MPa yield of forged 6061-T6 that gives a ~3.1× margin; even measured against the conservative 276 MPa standard-minimum yield for the alloy, the wheel still holds about ~2.7× — stricter than the 2× design factor common elsewhere. The point isn’t the headline material number; it’s that the margin survives the conservative assumption. Final structural validation and per-corner weight are confirmed before the wheel ships, the same way every LFI wheel is measured before machining.

As a Founder Edition, the REX-0R Superleggera is built to the specification Kevin would put on his own car — no compromise made to hit a number, every gram accounted for. It is forged to your width, offset and brake clearance, with no spacers and no shortcuts.