Welcome Guest, you are in: Login

FUBAR Labs Hackerspace Wiki



Search the wiki

Back to the Main Rocket Page


A single frame of my 3D printed engine from test2 on 2014-10-23

A single frame of my 3D printed engine from test2 on 2014-10-23

This page contains information related to my first attempt at building a liquid fueled regenerative 3D printed engine. Additionally, this is my first attempt at building an engine that is not also an igniter. In other words the engine has no ignition source so it has to be first started by an igniter.


The goal of this engine was to take what I’ve learned about regenerative cooling and apply it to a 3D printed engine.

I’m attracted to regenerative (and film) cooling as opposed to heat sink or ablative cooling because this approach could ideally produce an engine that can be fired more times with less structural wear from heat. I’m interested in 3D printing because it’s a useful tool for building complex objects that are not readily available

Printing Approaches

There are 2 main printing approaches for producing metal parts I looked at:

Details on each approach can be found in the aforementioned links but for purposes of this discussion DMLS is higher quality/resolution/cost and SLS is lower.

DMLS prices are significantly higher than SLS. Although, I suspect this will change in the next few years as 3D printing becomes more popular this factor alone forced me down the SLS path.


SLS is a 2 step process. The first step is to create a base structure with powdered 420 stainless steel. After this process it leaves a powder residue that must be removed manually via an air-hose before the object is then infused with bronze during the second step. The completed product has a brown color and usually a somewhat bumpy finish although it can be polished as part of a finishing process.

This 2 step process causes a problem for regenerative engines because the powder that is left after the first step can easily get trapped in the internal geometry of coolant passages. Then when the object is infused with bronze the powder hardens and blocks the passageways. As a result of this when 3D printing in SLS there needs to be entrance and exit holes designed so that air can be blown to remove the powder after the first step.

This last point is one of the hardest aspects about designing printed parts. Aside from this working with SLS is pretty straightforward. While it does not provide a dimensionally perfect part I do find it to be accurate enough** and the resolution (ie. level of detail) has so far been adequate.

**To provide a general sense of accuracy ExOne, which is one of the print services I use, guarantees parts to a +/-.5% of any given geometry.

Thermal Conductivity

A nice property of the bronze/stainless-steel SLS process is that *I believe* the bronze is giving the material a higher thermal conductivity then the engine would otherwise have if it were only made of stainless steel. This is a desirable property in a rocket engine as it helps with regenerative cooling.

I do not know the exact thermal properties of this material so it’s somewhat of a guess but from simple heating tests and during machining it feels like the heat is being distributed fairly quickly throughout the piece.


Machining SLS printed bronze/steel is difficult. The material is very hard and I’ve found that only cobalt steel or harder materials work. Even then machining operations are difficult and I’ve had problems with tools breaking.

I try to minimize machining operations to tapping and polishing areas that will mate. When designing threads for tapping operations I usually go to about 95-97% of design and then drill out the remaining material in order to get the correct hole dimensions

Trial & Error

Another aspect of printing to take into consideration is to plan to do at least 3 revisions before producing a satisfactory part. For example, if an engine costs $300.00 USD to print make sure your budget is for $900.00 USD. This is especially true if you are new to the process. 3D printing is like working with any other medium in that it takes some time to get comfortable with and understand its nuances.


The 3D printed engine is built in 2 pieces: an injector and the main engine. There is a flange that connects the 2 via 6 1/4"-20 screws and an O-ring seal.

CAD image of the injector and engine 2 piece design

CAD image of the injector and engine 2 piece design

Main Engine

The main engine was designed to be regenerative; however, in its current (second) iteration I’ve still not been able to successfully print the coolant channels. I believe this is mainly for the reasons stated previously regarding the de-powdering process after the first printing step. I’m somewhat confident I understand the process well enough that the 3rd version will successfully print coolant channels.

Had the regenerative cooling worked there would have been 10 tubular channels each with a 1mm diameter. The ethanol (fuel) coolant would enter the channels at the nozzle exit, circulate up to the top of the engine chamber where it would exit the engine and flow through another tube that connected to the injector. Having the coolant actually leave the engine before flowing into the injector via an external tube rather than flowing directly into the injector made the setup more modular in that I could design the 2 somewhat independent of each other.

Rather than continue iterating in order to fix the cooling I decided that the part looked good enough otherwise to test as a non-cooled engine.


Fuel and Oxidizer enter the injector through separate ports. Inside the injector are 2 tori. The inner one circulates the oxygen and the outer one circulates the fuel. The fuel and oxidizer are not mixed in the injector as per standard practice.

Fuel and oxygen are injected into the main engine chamber via a very simple unlike impingement configuration. There are 2 injector holes for the fuel and 3 for the oxidizer. They all impinge upon each other in the center of the engine, which is where the flame from the igniter is also injected. The 3 oxidizer each have 3/64" diameter orifices and the 2 fuel orifices are 0.04". Note that these orifices are larger than required so I use precision drilled orifices immediately upstream of the injector to finely control the flow rates.

clockwise from top: the top of the injector, bottom, and CAD design showing internal geometry

clockwise from top: the top of the injector, bottom, and CAD design showing internal geometry


In order to light the main engine I use my 3D Printed Igniter v3.

Screenshot of the 3D Printed Igniter testing

Screenshot of the 3D Printed Igniter testing


The engine plumbing consists of 4 propellant lines: 2 for fuel and 2 for oxygen. There are 2 lines for both fuel and oxidizer because one is for the igniter and 1 line for the injector. The igniter lines use solenoid valves (on/off) and the injector lines use servo valves (variable area). The variable area servo valves were used so that I could experiment with throttling however; I’ve had trouble accurately controlling flow via this method (see: Precise Flow Control. I decided to put this project on hold until I have the non-throttled engine working correctly.

image showing how the fuel is pressurized and pushed out of the tank.

image showing how the fuel is pressurized and pushed out of the tank.


For the oxygen line I use welding oxygen gas. I have a standard regulator, which is used to control the pressure and thus the flow of oxygen to the main engine. A pressure transducer sits upstream of the oxygen orifice and is used to calculate flow. All of my gas flows are choked so that I do not need to rely on downstream pressure to determine flow rates.


The fuel line is slightly more complex since the fuel is a liquid and thus must first either be pumped or pressurized before it will flow. I chose to pressurize the fuel in an aluminum paintball tank with the regulator removed. I chose a paintball tank because it is strong, rugged, small, affordable and easily obtained.

Fuel is first put into the tank via a funnel. Then a copper tube is inserted into the tank. The tank is then sealed and flipped upside-down.

The copper tube is used to feed an inert pressurizing gas (nitrogen) into the top of the tank to force the fuel down and out through a valve at the tank exit. In order to make this work the copper tube has a smaller diameter than the tank opening. This means that the pressurizing gas can be injected into the top of the tank thus forcing the fuel out of the tank and into the propellant lines.

Performance Characteristics

The engine was designed with the following characteristics.

  • Fuel Mass Flow: 0.0545 kg/sec
  • Oxidizer Mass Flow: 0.0545 kg/sec
  • Total Mass Flow: 0.1093 kg/sec
  • Design Mixture Ratio: 1:1
  • Design Force: 50 lbf
  • Design Chamber Pressure: 150 psia
  • Design Temp: 2572 Kelvin
  • Design Specific Impulse: 209 Isp


After designing the engine I realized that it was not possible to achieve my desired oxygen flow rate of 0.0545 kg/sec of gaseous oxygen due to some currently unknown limitation in my propellant feed system. Currently the maximum flow-rate is roughly 0.015 kg/sec at 300psia. I’m in the process of researching why this happening and to see if the flow rate can be increased.

In the mean time, however, I’ve had to reduce the total thrust of the engine significantly. This is discussed further in the testing section.


STL CAD files for the injector as well as the engine can be found on GitHub.

The aforementioned link also contains an Excel file showing the design and calculations. The file includes a diagram of the plumbing components as well. Note this file has been scanned for viruses before loading. There are no macros in this file.

In addition, there is a also a Rocket Propulsion Analysis (RPA) file showing the design parameters. There is some slight divergence between RPA and my Excel design most likely due to less accurate assumptions made in my analysis. Also, note that these calculations assume the higher oxidizer mass flow numbers and have not been revised to reflect the more limited flow rate.

Engine Firing Order of Operations

The ignition process is as follows:
  • -500 ms: start the spark plug igniter
  • -75 ms: open the GOX igniter valve
  • 0 ms: open the ethanol igniter valve
  • 1000 ms: run the igniter until this point to give the propellants time to come up to operating pressure.
    • Quickly open the main engine GOX valve and slowly open the main engine ethanol valve. This tries to prevent the ethanol valve opening too quickly and accidentally extinguishing the igniter flame.
  • 2000 ms: at this point the valves are fully open and the main engine runs.
    • Shut down the igniter ethanol and GOX valves
  • Cutoff time**: Stop the igniter, shut down the main engine ethanol valve quickly and the GOX valve slowly. This ensures that the flame burns as long as possible to prevent excess ethanol from pooling in the engine chamber. The GOX also acts as a sort of purge.
  • Test Termination: after cutoff, monitor the engine sensors until pressure reduces to ambient. Then complete the test.

** this is the number of ms the engine should run for and is provided by the user.

The below image depicts the process of testing an engine. The images are taken from test2 on 2014-10-23.

From top to bottom left to right: inital ignition, sustained ignition, opening main valves, main combustion starting, full thrust, and termination.

From top to bottom left to right: inital ignition, sustained ignition, opening main valves, main combustion starting, full thrust, and termination.

Test Results

Test results are below. Each test includes a description on the left and an image linking to a youtube video on the right. Click on the image to launch a video in a new tab.

The below image shows the layout of the engine on the test stand. This image was taken after firing on 2014-11-15.
Labled engine components while sitting on test stand

Labled engine components while sitting on test stand

2014-10-23: (Test2) First "Real" Engine Test

Still frame green (copper?) flame color and hot red throat section

Still frame green (copper?) flame color and hot red throat section

  • Test Results
    • Ox Orifice/Feed psia: 0.023”/281
    • Fuel Orifice/Feed psia: 0.113”**/204
    • Average Chamber psia: 26
    • Pre-Calculated Flow (kg/s) & OF Ratio: 0.0189, 1.77:1
    • (Estimated) Actual Flow (kg/s) & OF Ratio: 0.02, 1.44:1
    • Test Duration: 40000 ms
    • Test Video
  • Analysis
    • While I had done some very basic testing in late summer this is the first real test of my main engine. The results are far from ideal but none-the-less the results were within the margin of what I had anticipated.
    • The first thing to note is that the flame color has a bit of a green tint to it. From showing this video to a few people at FUBAR we came to the conclusion this is most likely copper (from the bronze infusion) burning. Notice how the flame starts blue and then turns green after about 2 seconds. The burning probably starts after the engine heats up.
    • Since the engine is a heat-sink it will continue to get hotter and hotter until it finally fails. Looking at the last few frames of the test we can see the throat area getting red as well as a section of the upper chamber. The hot (red) throat is not unexpected because the throat is the area of highest transfer. I have no explanation for why the top part of the chamber also turned red.
    • The startup process is as follows:
      • -250 ms: start the spark igniter
      • 0 ms: open the igniter valves
      • 1000 ms: slowly open the fuel valve and quickly open the ox valve. This is so that the fuel doesn't douse the igniter flame for any reason.
      • 2000 ms: main valves are fully open. Close igniter valves
      • 4000 ms: close main valves and shut off spark igniter.
    • Based off the above it's interesting to note that the fuel pressure takes about 150 ms before the pressure matches the main chamber. My hypothesis is that this is due to propellant still in the lines that has yet to reach the igniter. After about 150 ms the pressure goes down very quickly.
    • Initially the chamber pressure spikes to about 60 psia and then quickly goes back to around 25 psia. I've noticed this in tests for a few different engines/igniters and I'm not sure why its happening.

**The 0.113' orifice diameter is larger than the max orifice flow rate I could achieve during the test. So I scaled down the CD factor to 0.37 to compensate. The scaled down CD factor is based on flow testing analysis

ScrewTurn Wiki version Some of the icons created by FamFamFam.