It broke under my heel.
Now it is a branching aluminum part.
A printed heel guard keeps my boot off the chain on a 2001 Honda XR50. The first one snapped. I found the root cause, set up real loads for an adult rider, and let topology optimization find the load paths. That plastic part passed its fit check and is on the bike. Then I took the same load paths into metal: a twin-limb branching aluminum part, built from code, resized by FEA against eight load cases including crashes, printed in AlSi10Mg, painted to match the bike and bolted on.
The aluminum part on the bike. Real photo- Bike
- 2001 Honda XR50, stock frame
- Rider
- 180 to 220 lb adult
- Material
- AlSi10Mg, laser powder bed fusion, painted to match the bike
- Load cases
- 8: five riding, three crash, all under yield (simulated)
- Weight
- About 3.8 oz, 109 g
- Status
- Aluminum part painted and on the bike
Out of the box, painted, bolted on




Where it stands
The part came back from the printer bead blasted, got painted to match the bike, holes reamed to size, and went on with safety-wired bolts. It is on the bike now and holding up. Ride results and the proof test numbers land here as they come in.
Same load paths, a part that grows like a tree

Twin-limb branching arms
Every long load path is split into two limbs with cross links, like a tree branch. Each limb keeps one diameter along its length, so the part prints clean and the look stays consistent.
Branching faceplate
The heel pad is a recursive tree: four trunks from the center that fork twice on the way to the rim, with solid hubs where the arms land. It stops a boot heel with a fraction of the material of a solid plate.
Every member sized by FEA
A script maps the stress field back to each limb and resizes it against the allowable, then snaps the diameters to a few tiers. 12 g came off the previous version with the crash peaks held.
Built for the process
Bolt holes print undersize and get reamed. Mount and heel faces stay flat for support. M8 drilled hex bolts get safety wired, no washers. Bead blasted, then painted.



Eight load cases, all under yield

Pad toward the chain, peak stress, safety factor on yield
| Case | Pad | Peak | SF |
|---|---|---|---|
| Heel push225 lbf into the pad | 0.008 in | 13.2 ksi | 2.0 |
| Heel on outer edge180 lbf on the pad tip | 0.004 in | 7.8 ksi | 3.3 |
| Heel and scrub forward157 lbf in, 90 lbf forward | 0.002 in | 8.7 ksi | 3.0 |
| Heel and scrub back157 lbf in, 90 lbf back | 0.009 in | 13.6 ksi | 1.9 |
| Chain slap67 lbf outward from the chain | 0.002 in out | 3.9 ksi | 6.6 |
| Crash, tip-over450 lbf inboard on the pad tip | 0.004 in | 17.6 ksi | 1.5 |
| Crash, slide forward337 lbf in, 225 lbf forward | 0.005 in | 22.9 ksi | 1.1 |
| Crash, slide back337 lbf in, 225 lbf back | 0.009 in | 23.4 ksi | 1.1 |
- Riding loads clear yield by 1.9x or better. Crash cases are one-time events and clear by 1.1x or better: a local spot may yield, the part stays on the bike.
- The pad never moves more than 0.01 in toward the chain. The plastic part moved 0.05 in in its worst case, the original 0.22 in.
- About 3.8 oz. Less than an ounce over the plastic part, in aluminum.
Linear static FEA on the final print mesh. Yield taken as 26 ksi for stress-relieved AlSi10Mg, to be confirmed on the vendor sheet. A heel-stand proof test on the first part is still to come.


Six aluminum versions in a week

- V7.9
Plastic rebuild
Topology optimized PETG and ABS part. Printed, fit checked and on the bike.
- V8 to V8.2
First aluminum
Same load paths rebuilt as smooth round struts for metal printing, then a sculpted organic pass. Crash cases added.
- V9
Voronoi
Hollow oval tubes with a Voronoi window pattern. Strong, but 141 g. Too heavy.
- V10 and V10.1
Branching faceplate
Solid arms back, with a spoke and fork tree in the heel pad. 108 to 110 g.
- V11
Twin-limb arms
Every long arm split into two branching limbs with cross links. 121 g.
- V11.1
Resized by FEA
Each limb resized against its own stress. 109 g, all eight cases under yield. This is the one on the bike.
How the metal part was designed
- 01
Keep the load paths
The plastic part had already found where the material wants to be. The metal version keeps those paths and changes the material and the shape around them.
- 02
Add the crash cases
A tip-over onto the guard and a slide forward or back are what kill a part in the pits. 450 lbf on the pad tip and 337 lbf plus 225 lbf sideways joined the five riding loads.
- 03
Build shapes from code
Members, blends, branching rules and the faceplate tree are all parameters. A style change is a rerun, not a remodel. Six styles were screened before one was sized.
- 04
Mesh to FEA, FEA to mesh
Every build goes straight into a 1 mm voxel FEA. The stress field maps back to the members, and the members resize. Repeat until every case clears.
- 05
Check the print
Chain gap at the keep-out, support area heel face down, overhang on every limb, and a flat mounting face that bolts to the frame tab with socket access.
- 06
Print, paint, bolt on
Faceted STEP and a build-orientation STL to the printer. The part came back bead blasted, got painted to match the bike, holes reamed, and bolted on with safety-wired M8s.
Status: the aluminum part is painted and on the bike. Heel-stand proof test numbers and ride results go here as they come in.
A broken part is the best design brief
The problem
My pit bike runs a printed heel guard that bolts to the frame tab and keeps my boot off the chain. The first one snapped at the arm and cracked at the bolt holes.
The approach
I treated it like a customer failure. Root cause first, then load cases from how the part is really used, keep-outs for the chain, bolts and frame, and topology optimization with an FEA check on every version.
The result
A bridged, open-skeleton guard with a clean mounting face. In simulation it moves 3 to 8 times less toward the chain under the same heel loads, for less than an ounce more.
Where it failed and why


- 01A lightening window sat right where the bending load was highest.
- 02The print layers ran across that load instead of along it.
- 03The walls around the bolt holes were thin, so they cracked too.
Real loads in, only the useful material out


Three features that do the work

Bridge between the bolts
Both bolt bosses are tied together, so the load spreads across the frame tab instead of tearing at one hole.
Open skeleton heel pad
The pad only has to stop a boot heel. A rounded rim and spokes do that with less material, and give the supports a place to land.
Clean mounting face
Nothing sits behind the frame side of the part, so it bolts flat to the tab with clear socket access to both bolts.
Original to the plastic final

Stiffer where it counts

Pad movement toward the chain
| Load case | Original | New |
|---|---|---|
| Heel push, 225 lbf | 0.145 in | 0.043 in |
| Heel on outer edge, 180 lbf | 0.225 in | 0.028 in |
| Heel and scrub forward | 0.085 in | 0.017 in |
| Heel and scrub back | 0.218 in | 0.053 in |
- About 0.24 in of gap to the chain. The new part stays under 0.06 in in every case.
- Near-peak stress (99.5th percentile) drops from about 5,900 psi to about 1,600 psi in the worst case.
- 2.75 oz, up from 1.9 oz. Less than an ounce more for a part that is several times stiffer.
Linear FEA, same loads on both parts. The test print passed the fit check on the bike.
How the plastic part was designed
- 01
Failure review
Found the cause: a window in the highest bending zone, layers across the load and thin walls at the holes.
- 02
Real loads
Heel push, heel on the edge, scrub both ways and chain slap for a 180 to 220 lb rider.
- 03
Keep-outs
Chain in position with clearance, socket access to both bolts and a clean mounting face.
- 04
Topology optimization
Started from a block of design space and kept only what carries load toward the frame.
- 05
Iterations
Bridge, rounded pad, open skeleton, clean mount. Each change came from a real problem on the bike.
- 06
Smooth and check
Smoothed into a printable part and checked with FEA every time it changed.
The test print passed the fit check on the bike and the plastic part runs at the late October moto events. Everything it learned about load paths went straight into the aluminum part above.
What this means for your project
Root cause first, then a better part.
Send me the broken part or photos of it. I find out why it failed, build load cases from how it is really used, and redesign it so the fix holds up.
Lighter, stiffer and built to be made.
Topology optimization and FEA to set the shape, then a design that fits the process, whether that is a plastic print, metal AM, a machined part or sheet metal. You get native CAD, STEP and drawings.
Got a part that keeps breaking?
Send me photos and what it has to do. I review every request myself and reply within 2 business days.
Start a project→