SUPERVISOR
Arya Dehghani
ROLE
Responsible Engineer
Hotfire Test Director Feed System Design
Test Stand CAD
Pressure & Thermal Analysis
Fabrication
DATE
Sept. 2025 – Present
The Prometheus System is a reusable torch igniter which employs a flame to light liquid fuel rocket engines. It taps off the engine’s feed system, Atlas, redirecting a small amount of propellant through orifices and into the torch chamber, igniting it with an augmented spark plug and circuit.
It employs a heat exchanger to vaporize liquid to gaseous oxygen, enhancing mixing. The torch has a dedicated test-stand feed system designed to replicate the pressures seen on Atlas. The test stand allows us to quantify ignition repeatability, propellant mixing, and spark circuit reliability before engine integration.
Purpose & Scope
My team and I are built this system from scratch, as it's previous iteration encountered major pressure losses and unpredictable flame behavior. The system is to be integrated into the teams current Nomad Engine and Ranger flight vehicle, tapping off the Atlas feed system.
With pressure drop calculations using traditional and modified Darcy-Weissbach equations, It was found that with an assumed oxygen tank operating pressure of 663 psi and a fuel tank operating pressure of 995psi, the feed system would see a pressure drop of no more than 10 psi in the compressible GOx line – these being conservative estimates. This is a massive improvement over the first iteration of the system which saw pressure losses of well over 200psi.
During P&ID design, I ensured all actuated solenoid valves are complemented by manual valves incase of failure. The manual valves are oriented to operable from in behind the test stand with a system diagram etched onto the backside of the test-bed plate, providing easy access in rare cases where they may need to be used by operators – for example during solenoid failure.
Despite gaseous oxygen, we can assume incompressible conditions due to a gas velocity of under Mach 0.3. These values are well within the required range for reliable operation, which in our case requires a pressure of roughly 600psi from each line to the torch. Our new system was validated and we moved onto stage 2 of design.
Atlas-Integrated System
Test System CAD
Integrated Supply Line and Pressure Regulators CAD
Integrated Heat Exchanger CAD
Integrated Torch CAD
Feed System Design
The idea behind the initial torch ignition system's CAD was to create a test bed where we could fire the torch, gather data, and check for losses, leaks, and potential improvements before building the final system. Importance was placed upon system safety during operation and ease of access to all valves and ports. I opted for a symmetrical design, attempting to have the oxygen and fuel mirror each other where possible. We selected Jet-A (refined kerosene) as the fuel source for simple and reliable ignition.
The fittings were AN / NPT / Swagelok, and all piping to be 1/4 inch. We inserted check valves to solidify flow direction in case of back-pressure, micro-particle filters after tanks to protect components, and pressure release valves to ensure an active pressure ceiling during operation. The piping presence is kept to a minimum, reducing bends and pipe length where possible to minimize losses. This further creates a simple test bed where troubleshooting is easily completed. Pressure transducers capture readings throughout, and a thermocouple was secured to the torch's exterior surface to ensure the o-rings used to seal fittings remained under 500F (within operating conditions) The piping was secured to the test stand's 30" x 30" x 1/8" aluminum sheet with 3d printed tube clamp mounts.
The solenoids are a combination of normally closed (N.C) inline valves and normally open (N.O) vent valves to ensure safety during power loss.
A Nitrogen tank is used to pressurize the fuel tank (TNK-F) and purge the feed system and torch of any ignitable particles or fuel (FOD, IPA, kerosene). FLT-F/N/O are filters which further remove any contaminants that may be present in the tanks.
Part selection and a bill of materials for the test bed and torch were completed – keeping entire system cost under $4,000.
Bill of Materials. (Swipe for part selection examples)
Reused Tank
Reused PRV @ 1200psig and new PRV @ 1000psig
New ‘Normally Open’ solenoid valves
Torch Design
- The mixing mechanism was chosen to be coaxial swirl. Initially we were aiming to have the torch operate based on a coaxial shear mechanism but press-fitting the fuel post, which would operate as an injector, was not structurally possible due it's small size and inevitable deformation, along with the excessive forces required to press it through the torch body. Instead we opted for NPT threaded orifices for easier installation/removal. A PT inlet (5/16”-24 ORB) was added to measure torch chamber pressure for our research on glow vs spark assisted igntion.
This work is presented below along with Isai Andrade's CAD of our torch, it's requirements, the material selection trade study, manufacturing plan, structural and thermal analysis.
The slides presented have been extracted from out team's Successful CDR/TRR
Swipe for more images
Spark Plug Torch Design
Glow Plug Torch Design
Spark Circuit
Alongside this, the team is has finished the igniter circuit design (spark to light the torch) which is proposed to operate at 30kV with an input of just under 5V. It is to be manufactured in house. The design and part selection was undertaken by Faith Colon. The Enclosure for the exciter and circuit was created by Adrian Tejada. Their work is presented below.
Swipe for Images
Fabrication
A component pressurization rig was created to test for solenoid leaks since we reuse these valves. For the test bed, 1/4" OD lines were flared or swaged accordingly and hydrostatic proofed to 1500psi (1.5x MEOP). They were then ox-cleaned according to ASTM G93 standard with simple-green and IPA and sonicated. This was done to the torch as well. The torch was machined using MasterCAM and a 3-axis CNC.
Cold Flows
Using GN2 and water to represent GOx and kerosene, we leak checked and flowed through the test stand to characterize pressure transients in the torch. This was done so we could complete our ignition sequencing. The Initial transient flow is when we decided ur spark signal is sent and ignition will occur. The cold flows also helped us determine more accurate regulator set pressures to achieve a fuel and oxygen MR = 1.5. Our pressure plots and sequencing is shown below.
Test Campaign
On May 2nd, 2026, team Prometheus successfully hotfired 17 times.
We ignited both our glow and spark-assisted torches, achieving chamber pressures of over 470psi and sustaining a 2 second glow-assisted hotfire. The spark-plug torch was fired 10 times without reapproaching the pad. The minimum chamber pressure requirement was met and our repeatability goal of 90% or more was achieved.
This is a description underneath your image.
Heat Exchanger
While we are tapping into liquid filled tanks for the ignition fuel/oxygen sources, our torch will operate under gaseous Oxygen settings and liquid fuel settings. This is achieved by increasing the temperature of the oxygen with a heat exchanger in the form of coiled tubing, aiming for temperatures at or close to 300K to ensure the vales don't freeze. We are therefore removing the need for cryogenic compatible system components, further reducing cost. On top of this, using GOx has the advantage of giving us predictable flow behavior. We are far less likely to encounter flashing or 2 phase choking anywhere near the torch, and more likely to have a stable energy of reaction.
With a gaseous Oxygen mass flow rate of 3g/s, at pressure 620psi and temperature 300K, we found the volume of gas required for a 1 second flow (per ignition) to be just under 3.5 cubic inches. To accommodate this volume, I designed the heat exchanger to be 10 feet long with inner diameter 0.19 inches a coiling diameter of 3 inches. LOx enters and exits as GOx. This means the heat exchanger can also function as an accumulator and allow us to visually see the LOx to GOx transition according to the frost line. This oxygen volume allowed us to eliminate the need for a secondary tank on the test stand. Simplifications and cost reductions such as removing the need for an oxygen tank carry the bulk of our design choices.
A thermal analysis of the heat exchanger was completed and we found that when the exchanger was filled entirely with LOx at -160 Celsius, that mass of Oxygen would vaporize fully to GOx in under 8 seconds. To complete this, the exterior natural heat transfer coefficient was found, along with the conduction in the copper tubing and convection on the interior of the tube. Hand calculations and an Ansys simulation were completed to verify.
This part is not included on the test stand as we are testing with GOx to simulate a main feed system tap-off.
Heat Exchanger CAD