3D Printed DBR22 Sports Car: How the Rear Structure Works

• 1
  What the 3D printed DBR22 sports car rear structure actually is
• 2
  Why print the nodes instead of machining them
• 3
  How bonded joints behave under cornering and heat
• 4
  Where CNC machining still takes the load
• 5
  Material choice for printed and machined rear parts
• 6
  Inspection and fit-up for a bonded structure
• 7
  Printed node vs machined part: which process fits

Start with the car. The DBR22 is a two-seat speedster: 705 hp from a twin-turbo V12, no roof, no fixed rear glass. Removing the roof takes away a load path. On a coupe, the roof and rear bulkhead tie the two shock towers together, so cornering loads travel through the shell. On an open car, that job moves down into the rear subframe and the body structure behind the seats.

That is the part printed. The rear support structure is not one big printed block. It is a set of aluminum nodes, each printed to near-net shape, then bonded into an assembly. Think of a space frame built from a few thick junctions instead of dozens of welded tubes. The nodes carry the concentrated loads; the tubes and panels between them carry the bending.

Why aluminum. The V12 sits behind the driver, so the structure around it sees heat, vibration, and cornering load at the same time. Aluminum keeps mass low, and printed aluminum lets the designer thicken only the walls that see stress. A machined billet node would need the same wall thickness everywhere because the tool cannot reach inside a closed pocket.

One number worth keeping in mind: the whole point of printing here is stiffness per kilogram, not peak strength. The structure has to hold the rear body, the exhaust, and the suspension pickups without moving more than a few tenths of a millimeter under cornering load.

• 1
  Printed nodesNear-net aluminum junctions, thickened only where stress runs
• 2
  Bonded assemblyAdhesive joins nodes to tubes and panels, no weld distortion
• 3
  Load pathCornering load moves from the shock towers into the rear subframe

A printed node wins for one reason: internal geometry. A node often needs a hollow core, a tapered wall, or a rib pattern that follows the load. A 5-axis milling cutter enters from outside, so any cavity it cuts must be reachable and must drain chips. A printed node has no such rule. Walls can start thick at the flange and thin out toward the middle.

The second reason is lead time at low volume. A concept car runs in single digits. Printing a node takes hours of machine time but almost no setup. Milling the same node needs fixtures, soft jaws, and a first-article check before the second part is safe to cut. At one or two pieces, printing usually reaches a usable part faster.

The third reason is weight. Printing lets the wall thickness vary along the part. On a machined version, the minimum wall is set by the cutter and the clamping force, so the whole node ends up heavier than it needs to be. On a 705 hp car with a light rear end, that extra mass sits in the worst place.

Where printing loses: surface finish and tolerance on mating faces. A printed node lands around Ra 6–12 μm as-built with visible layer lines. Any face that bolts to a machined pickup, a bearing bore, or a threaded insert still needs cutting. That is the normal split on these cars: print the node, machine the interface.

• 1
  Hollow coresPrinted walls can thin out internally; a cutter cannot reach inside
• 2
  Low volumeNo fixtures needed for a one-off node
• 3
  Finish limitAs-built surfaces need machining at bolted and bearing interfaces

An adhesive joint does not fail the way a weld does. A weld has a heat-affected zone and a stress concentration at the toe. A bonded lap joint spreads load across the whole overlap area, so peak stress at the edge drops. That is why bonded aluminum structures tolerate fatigue better than spot-welded ones at the same thickness.

The catch is peel. Adhesives are strong in shear and weak in peel, meaning any load that tries to pry the joint open at the edge is the dangerous one. Designers fight this by adding a flange lip, a mechanical interlock, or a rivet at the corner so the joint never sees pure peel. If you see small fasteners near the ends of a bonded joint on a car like this, that is usually why.

Heat changes the story. Underhood air near a twin-turbo V12 can sit well above ambient, and most structural adhesives lose modulus as temperature climbs. The joint does not fall apart, but it moves more. That movement shows up as a change in suspension pickup location, which the driver feels as a shift in steering.

Bonded joints also need clean surfaces. Anodized or primed aluminum bonds well; bare aluminum with cutting fluid residue does not. On a machined mating face, the shop has to specify the cleaning step, not just the dimension. This is a process control issue as much as a design issue.

• 1
  Shear vs peelDesign so the joint sees shear; add a lip or rivet against peel
• 2
  HeatAdhesive modulus drops with temperature, so pickup points move
• 3
  Surface prepContaminated aluminum bonds poorly even with the right adhesive

Not every part in a structure like this should be printed. Machining holds the tolerances that make the assembly work. A printed node can be off by a few tenths of a millimeter and still bond fine, because the adhesive gap absorbs it. A suspension pickup point or a bearing bore cannot. Those need ±0.005 mm, and that comes off a mill or a lathe.

Machined parts also carry threads better. A printed aluminum thread is weak in shear and strips easily at the pitches used on automotive fasteners. The usual answer is a machined insert or a machined boss that the printed node bonds around. GreatLight machines those interfaces on 5-axis centers with a Ø400 mm rotary table, which keeps the bolt pattern and the bore in one setup.

Rigid parts are another case. If a part is a simple plate, a bracket, or a shaft with no internal cavity, printing adds cost and surface cleanup for no benefit. Cutting it from 6061-T6 or 7075 billet is faster and gives a better surface. The break-even is not about the technology; it is about whether the geometry has a reason to be hollow.

Finally there is repair and rework. A machined interface can be re-cut if a dimension drifts. A printed node with a bad face often has to be scrapped. On a concept car where the schedule matters more than the part cost, that difference decides the plan.

• 1
  TolerancePickup points and bores need ±0.005 mm; print cannot hold it
• 2
  ThreadsMachined bosses and inserts take fastener load
• 3
  Simple shapesPlates and brackets are cheaper cut from billet

Printed aluminum for automotive nodes is usually a casting-grade alloy such as AlSi10Mg or a scandium-modified grade. These print cleanly, weld well, and hold up under vibration. They are not as strong as 7075, but a node is not a highly stressed part; it is a load distributor. The strength lives in the tubes and the bonded overlap.

Machined parts near the same structure use different grades. Suspension arms, uprights, and engine brackets typically run 6061-T6 or 7075 for strength-to-weight. If the part sees heat near the exhaust, 7075 loses strength faster than 6061, so the choice moves back to 6061 or to a stainless grade. GreatLight stocks 6061, 6061-T6, 2024, 5052, 5083, 6063, 6082, 7075, and ADC12, plus 303, 304, 316, 316L, 17-4PH stainless when corrosion matters more than mass.

Titanium is the third option. TC4 (Ti-6Al-4V) machines slowly and costs more per kilogram, but it holds strength at exhaust temperatures where aluminum has already softened. Use it for a small bracket near the turbo, not for the whole rear frame.

Do not pick a material from a data sheet alone. A printed node and a machined insert in the same assembly should be checked for galvanic pairing and for thermal expansion mismatch. Aluminum against titanium moves differently as the exhaust heats, and a rigid bonded joint between them will feel that difference.

• 1
  PrintedAlSi10Mg or Sc-modified grades for nodes and load distributors
• 2
  Machined6061-T6 and 7075 for arms, uprights, and brackets
• 3
  Hot zonesTitanium TC4 near the turbo where aluminum softens

A bonded structure has no weld to inspect, so the checks move to the interface. The first check is fit-up: measure the gap at each joint before adhesive is applied. A gap that is too tight starves the joint; too wide and the adhesive cannot bridge it. On a printed node, that gap is set by the printed face plus the machined face, so both matter.

The second check is the machined face itself. Bore diameter, bolt pattern position, and flatness get measured and recorded. GreatLight inspects 100% of parts before shipment and can supply reports on request, covering raw material check, in-process monitoring, and final inspection. For a rear structure, that record is what tells you the assembly will line up on the car.

The third check is after bonding. A tap test or a simple ultrasonic scan finds voids and disbonds. It is not a full NDT program, but it catches the joint that was never going to carry load. Print the node, machine the face, bond the assembly, then inspect the bond. Each step has its own failure mode, so each step gets its own check.

One practical note on tolerance stack-up. A printed node at ±0.2 mm plus a machined face at ±0.005 mm plus adhesive thickness adds up across a multi-node assembly. Do the stack-up before you release the drawing. If the total drift is larger than the pickup point allowance, tighten the machined face, not the printed one; it is cheaper to hold.

• 1
  Fit-upMeasure adhesive gap at every joint before bonding
• 2
  Face checkBore, bolt pattern, and flatness recorded per part
• 3
  After bondTap test or ultrasonic scan for voids and disbonds