Project Initium - Liquid Rocket Engine

Initium was a project I started and led in my last semester at Pitt and then continued to work on post-graduation with the rest of the engineering team. This project had the goal of designing, manufacturing, and testing Pitt's first ever student-developed liquid rocket engine using isopropanol and nitrous oxide as the fuel and oxidizer, respectively. The engine system was successfully Cold Flow and Hot Fire tested in April 2026.

On this project, I served as the Chief Fluid Systems & Operations Engineer on a team of eight other students. In my role, I was responsible for: 

1. Designing and assembling a floating-piston style propellant tank to safely store and feed the IPA and N2O propellants at pressures up to 1,000 psi.

2. Engineering a piping system to remotely fill the N2O tank and feed the propellants to the injector assembly at an expected flow rate of 1.1 lb/s and pressures exceeding 800 psi.

3. Sensor device selection to reliably gather critical pressure and temperature data within the propellant tank, piping system, and injector manifold.

4. Safe operating procedures for safely testing the engine system while minimizing risks associated with high-pressure systems.

Videos

Piping System Design Overview

The above P&ID was created in AutoCAD, and it shows the as-built version of the piping and instrumentation system.

The system incorporates three main safety devices to prevent over-pressurization when the fluid system is active. The main system is the static vent on the N2O-side of the tank, which allows the system to fill with liquid and, more importantly, vent off pressure even in the event of a power or communication failure. Additionally, there are two pressure relief valves set to 1,200 psi that can independently vent either propellant during a high-pressure event.

In addition to safety, electronic devices were sourced and installed in the propellant tank and piping system for system monitoring and remote control. On the Initium system, four pressure sensors and one temperature sensor (RTD) are installed in the tank and piping to analyze tank pressures, feed pressures, and pressure drops across the system. Additionally, every ball valve is servo-actuated to allow for remote control of the entire feed system, which minimizes the amount of time that a person must spend near pressurized equipment. 

The system also incorporates many passive instruments to maximize reliability and performance when in operation. Both propellant feed lines contain a Y-strainer to prevent debris from clogging the sensitive injector orifices. Past the strainers, the check valves on the fill and feed lines prevent back-flow and keep fluid flowing in one direction. Finally, flexible braided hoses are utilized for vibration-resistance and quick connection points during assembly. 

The 3D modelling and assembly for Initium was done primarily in a collaborative OnShape project space. On this system, I laid out the fluid system hardware in the test stand assembly to gauge system layouts, dimensions, and possible constraints or red flags prior to physical construction. The image above shows the final 3D assembly of the engine fluid system from which the physical Initium system was based on. 

Tank Hardware Design Overview

Initium's propellant tank stores the fuel and oxidizer inside a single structure built from a 3-inch Sch. 40 aluminum pipe. Inside the tank is an aluminum floating piston, which separates the two propellants to create a stacked-tank configuration. 

The floating-piston design employed by the tank was selected to utilize the self-pressurizing properties of nitrous oxide. At a temperature of 70°F, N2O has a saturation pressure of 760 psi. This high pressure serves as the driving force for fluid delivery, and the internal floating piston is used to pressurize the IPA and N2O liquid. 

This tank configuration removes the need for an external pressure source, such as an inert gas or pump. However, this also creates the risk of the propellants mixing inside the tank before reaching the combustion chamber, which could be an explosion hazard. Careful calculations and testing were performed on the o-ring glands for the floating piston to ensure that a proper dynamic seal was achieved as the piston was pushed upward.

The tank successfully held pressure during two engine tests at a peak pressure of 880 psi for 40 seconds. Post-test analysis revealed there were no leaks on any sealed interface, and all the hardware remained structurally sound and ready for re-use.

Although the propellant tank wall was able to be purchased as an off-the-shelf part, each bulkhead was custom designed and machined out of aluminum for this project. 

Each bulkhead has three o-ring glands as a critical redundancy to ensure no propellants leak out of the tank from the end bulkheads or mix by leaking past the floating piston. Every o-ring gland in this tank is sized to house a -232 Buna-N o-ring with 15% compression.

The two ends of the tank (IPA and N2O bulkheads) are retained using eight radial 1/4"-20 bolts. These bolts pass through clearance holes in the tank wall and thread into the bulkhead body to contain them inside the tanks. Assuming a working pressure of 800 psi inside the tank, these radial bolts operated with a factor of safety of 3.2. 

The IPA bulkhead (left image) contains two threaded NPT ports, one for the discharge/pressure relief piping and one for a pressure sensor. 

The floating piston (center image) is a hollow puck structure that separates the two propellants and transfers the N2O pressure to the IPA. The N2O-facing side of the piston is hollowed out to create an ullage space for N2O vapor to compress, which reduces the risk of hydraulic lock and over-pressurization.

The N2O bulkhead (right image) is the most complex with four threaded NPT ports for the filling/feeding piping, pressure sensor & pressure relief piping, RTD temperature sensor probe, and 0.025" orifice static vent structure. 

As mentioned above, the static vent structure serves two primary functions. The main function is to allow the propellant tank to fill with N2O liquid from the supply bottle. Because N2O has a very high vapor pressure, simply connecting the supply bottle to the propellant tank would cause the propellant tank to become pressurized with N2O vapor and prevent liquid flow once the pressure is equalized. The static vent utilizes an internal dip tube to vent vapor from the N2O headspace so the volume may be replaced with liquid. A 0.025" orifice is used to control the venting rate of the vapor and thus the inlet flow rate of the liquid. 

The second purpose of the vent is to provide a constant flow path for N2O pressure relief. If the electronics that control valve actuation were to fail due to power or communication loss, the static vent ensures that the system pressure can be safely relieved passively. 

Thermodynamic Model

To estimate the conditions inside the tank in preparation for the Cold Flow and Test Fire engine tests, I created a thermodynamic model in Python for this specific system.

The model runs a simulation for 60 seconds with 0.050-second timesteps to predict the steady-state pressure inside the propellant tank right before firing. The pressure and fill time outputs were the most useful for predicting estimated fill times and head pressure in the piping system. Several assumptions and simplifications were used in this model to simplify the complex behavior of nitrous oxide at our operating conditions. Such simplifications include:

1. Saturated Vapor-Liquid Behavior - Tabulated saturated liquid and vapor phase properties from NIST are referenced by this model to perform calculations. This assumption can be valid during the pseudo steady-state portion of the fill (where the graph plateaus) because the pressure inside the tank remains relatively constant. However, this means the model cannot predict transient behavior, including the initial pressure ramp. For the purpose of this simulation (to assess the working pressure inside the tank), it still provides useful information.

2. Choked Vapor Flow Through Static Vent - Because this model assumes a near instant pressurization inside the tank to keep the Saturated Behavior assumption valid, it is always assumed that the flow through the 0.025" static vent orifice is choked. This assumption is a realistic one due to the high-pressure differential between the static vent (600-900 psi) and the ambient pressure it vents to. Additionally, this model assumes only vapor is vented through the vent, even though a vapor-liquid mixture may be vented at times. This assumption assumes that the Fill valve closes instantly when the tank is liquid-full, which does not happen in reality. However, to model the steady-state behavior in the tank, this assumption remains realistic during the majority of the fill phase. 

3. Constant Supply Bottle Pressure - the K-bottle used to fill the tank is modeled at a fixed pressure throughout the whole process. In reality as liquid discharges the bottle cools slightly, dropping its vapor pressure over time. For our large K-bottle over a 30-second fill this drop is small and this pressure change is negligible.

Despite these simplifications, the model proved to be useful in estimating expected fill times and operating pressures prior to running the critical flow tests. The Cold Flow and Hot Fire ambient and K-bottle conditions were used to generate the two simulations below.

Testing & Results

The data above shows the system pressures for the Cold Flow test on 4/4/2026. IPA pressure, N2O pressure, and Injector Manifold pressure are displayed on the left y-axis. Valve states for the IPA Feed valve, Oxidizer Feed valve, and Fill valve are shown on the right y-axis. 

The Cold Flow data served to provide a real-world baseline for expected operating conditions in preparation for the Hot Fire as well as to assess the validity of the team's thermodynamic models. When comparing the tank fill simulation to the real Cold Flow data, the results are as follows:

• Steady-state filling pressure is overpredicted by ~40 psi (4%)

• Fill time is overpredicted by ~6 seconds (20%)

The data above shows the system pressures for the Hot Fire test on 4/12/2026. IPA pressure, N2O pressure, Injector Manifold pressure, and Combustion Chamber pressure are displayed on the left y-axis. Valve states for the IPA Feed valve, Oxidizer Feed valve, and Fill valve are shown on the right y-axis. 

As seen in the data, instability in the system caused significant oscillations in the operating conditions, likely caused by combustion instability after ignition. This instability propagated backwards, affecting all data streams. Despite this, the engine still produced a peak thrust of 105 lbf (470 N) and burned for approximately 3 seconds.

When comparing the tank fill simulation to the real Hot Fire data, the results are as follows:

• Steady-state filling pressure is underpredicted by ~40 psi (4%)

• Fill time is underpredicted by ~6 seconds (14%)

Comparing the two simulations to the two filling tests performed on the tank, it can be deduced that the model can predict the steady-state pressure in the propellant tank with ~90% accuracy. The fill time prediction sees larger deviations, however the simulation still provides good "ball-park" estimates of fill duration prior to firing. This estimate was also not used to gauge the N2O liquid level inside the tank. As mentioned previously, the presence of liquid being vented from the static vent was the indication used for determining when to fire. 

Assembly Images