HOWTO

 

Printed Components

It's clear that additive manufacturing (3D printing) is the future of short-run fabrication. However, it's far from clear whether the mechanical properties of typical 3D printed parts are sufficient for high-power rocket construction. This article documents my experiments get some numbers by testing three hypotheses:

  1. 3D printed parts are weaker than those made of traditional materials.
  2. Traditional materials produce parts that are stronger than they need to be.
  3. By optimizing the material, design and printing techniques, we can make printed parts strong enough.

Clearly these hypotheses are somewhat contingent on each other, but I'm testing the first two in parallel because I think both are interesting in themselves. Of course, proving the third is the ultimate goal.

 

The Tests

The first round of tests includes four different structures to test the absolute strength of ¼" (6mm) plywood centering rings (CRs) and bulkheads and the strength of a typical epoxied assemblies using them. I chose the most common HPR size: 98mm (3.9") body tubes and the CRs are for 54mm MMTs.

Test 01   Test 02

Break CR by pushing down on center.

Test the absolute strength of a CR by pushing a lip on the inside against a lip on the outside.

Break epoxy bond by pushing on center.

Test the weakest component of a bonded CR by pushing a lip on the inside of the CR.
Test 03 Test 04

Break bulkhead by pulling up on center.

Test the absolute strength of a bulkhead by pulling a ¼" bolt/fender washer through a hole in the center against a lip on the outside.

Break epoxy bond by pulling on center.

Test the weakest component of a bonded bulkhead by pulling a ¼" bolt/fender washer through a hole in the center.

Tests 01 and 02 simulate motor thrust being transferred to the body tube through the CRs. Tests 03 and 04 simulate parachute opening force where it's tethered to a bulkhead. Note that there are two series for Test 04: without and with and reinforcing coupler ring above the bulkhead.

The baseline parts are ¼" 5-ply "birch" plywood and the body tubes are Giant Leap flexible phenolic. All parts are for standard 3.9" (98mm) ID tubing. Epoxy bonds were AeroPoxy ES6209 with "gloved finger" fillets.

 

The Samples

The baseline material was ¼" 5-ply "birch" plywood, and the 3D printed samples were various materials and printers. Note that the filament was purchased from the printer manufacturer and the prints were all run with default printer settings.

Material Printer Technology Series
OnyxMarkforged MarkTwoFFF10
Onyx + CFMarkforged MarkTwoFFF11
ABSUltimaker S5FFF20
PETGUltimaker S5FFF21
PLAUltimaker S5FFF22
Tough PLAUltimaker S5FFF23
ABSBambuLabs X1CFFF30
PAHT-CFBambuLabs X1CFFF31
PETG-CFBambuLabs X1CFFF32
PCBambuLabs X1CFFF33
PLABambuLabs X1CFFF34
Tough 2000FormLabs Form3resin40
PA12HP Multi-Jet FusionFDM50

Note that all of these are FFF printers except for the Form3 which is resin and the HP MJF which is FDM. The HP MJF parts were purchased through Shapeways; the other parts were from printers I own.

Three samples of each series were tested in order to guage the variance in material and fabrication. (PLA was not used for CRs because of its low heat tolerance.) Each sample is identified with three numbers, for example "01-20-03" is:

Here are some of the samples for tests 01 and 03 and the aluminum fixtures used for testing them.

 

Hypothesis 1

3D printed parts are weaker than those made of traditional materials.

This turned out to be very much of a mixed bag. Some 3D printed parts (even in plastic) were much stronger than plywood, while others were much weaker. This proves a modified form of this hypothesis:

Parts printed on consumer-grade printers with default settings are much weaker than plywood.

I want to stress the "with default settings" here. These numbers should not be read as what these machines are capable of, but only what they produce by default. You can compare filaments on the same machine, but not between machines.

SeriesMaterialMachineForce (lbf)
01-01plywoodShopBot771
01-10OnyxMarkTwo433
01-11Onyx+CFMarkTwo1327
01-20ABSS5406
01-21PETGS5497
01-30ABSX1C260
01-31PAHT-CFX1C393
01-32PETG-CFX1C214
01-33PCX1C429
01-40Tough 2000Form31294
01-50PA12HP MJF594

Of note: parts printed with continuous CF reinforcement on a MarkForged MarkTwo or using Tough 2000 resin on a Form3 are almost twice as strong as the equivalent plywood parts. All other samples were significantly weaker.

I want to stress the "with default settings" again. You can compare filaments on the same machine, but not between machines.

 

Hypothesis 2

Traditional materials produce parts that are stronger than they need to be.

The first of results I got was for all four tests with plywood and phenolic, which brought this hypothesis into doubt. The breaking force of the plywood centering rings and bulkheads was not higher than the breaking force of the bonded assemblies. Below you can see that the parts and assemblies broke at roughly the same force:

Component (lbf)Assembly (lbf)
CR771799
bulkhead616611

Note that in the assemblies, the glue bond did not fail. Instead the ring crushed and the tube buckled. So the epoxy bond is also not the weakest link; it seems that all parts are roughly equally balanced for strength.

The conclusion I'm taking from this test is that comparing the strength of 3D printed CRs to plywood CRs is a reasonable proxy for "strong enough" (not over-building relative to the other materials and techniques).

 

Hypothesis 3

By optimizing the material, design and printing techniques, we can make printed parts strong enough.

This is the experimental part of the job: What geometry and printer settings will increase the strength of a consumer-grade printed part to that of plywood?

 

Raw Data

Here is the raw data from the testing lab. Note that I broke three samples of each unique combination and the values displayed above are the average of those three samples. (Force is in lbf and displacement is in inches.)

 

Test 01

Each test result: Test 01 spreadsheet.