Lotus Liquid Engine Cold Flow #1

Dec. 5 2015

BURPG has completed the first successful cold flow of the Lotus liquid engine, obtaining valuable data on the new ground fluids system performance, valve timings, and preliminary engine parameters. It was a significant first step in the development of the Lotus engine and further progress on BURPG’s way to space.

The testing included 4 oxidizer flows and 3 fuel flows. These varied in length, with the longer flows providing data on engine behavior, and the shorter flows providing a larger data set on valve timings. Determining these timings are critical for the engine ignition autosequences. Following is a brief analysis of our results and how these will be applied to future testing and revision.

See the video of the test here:  https://youtu.be/oEC2i0adbSg


Pressure drops are important engine parameters that can be found during cold flow testing. The lotus pressure drop data provides some interesting insight into the engine behavior during the cold flow.

Below is a graph of the fuel system pressure drops from a long duration fuel flow. The regen dP is the pressure drop across the regenative cooling channels, and is determined by the difference between the regen manifold pressure and the fuel injector manifold pressure. Other pressure drops are determined by measuring the pressure relative to atmosphere. Before the main fuel valve (MFV) opens, there is pressure present from the purge system. The purge check valve opening is also responsible for the small spike after the main fuel valve closes.

For the regen pressure drop, we were expecting 200psi, and the fuel injector 60psi. The low values shown in the graph are caused by higher than expected losses on the ground fuel system, and can be corrected for by increasing the test tank pressures. The mass flow rates that were calculated also came in at about 75% of what was expected, in agreeance with the lower than expected pressure drops.

The graph below details oxidizer injector pressure drop. As noted with the fuel, the purge pressure is seen again, this time as the pressure in the regen system and fuel injector. Purge pressure in the fuel system increases during oxidizer flow as the check valve for oxidizer purge shuts, diverting more flow to the fuel purge.

For the oxidizer injector, the expected pressure drop of 80 psi is actually lower than measured. The oxidizer mass flow rate was also calculated to be about 37% of what was expected. While this seems counterintuitive to have lower than expected flow rates and higher pressure, this can be explained. During cold flow conditions, the vapor pressure of nitrous oxide is significantly higher than the chamber pressure, causing the liquid nitrous oxide to cavitate and turn to gas in the injector ports, restricting the flow area and leading to the behavior seen in the data. Since this data does not verify the pressure drops of the oxidizer injector, other methods can be used. The CdA of the injector can be verified through water flow testing, from which the approximate pressure drop can be backed out.

Visible in both graphs is the effect of the pressure swings from the pressurization system, creating fluctuations in regen pressure and oxidizer injector pressure. More discussion on the ground support equipment performance follows.



Since we were using an entirely new ground support equipment (GSE) system for the liquid engine, the cold flow was as much a test of the GSE as the engine itself. The GSE system performed well for a first test, but observations and data from the test highlight some areas to focus on to make the system even more effective.

One area is the pressurization system. The press system is intended to keep the test tanks within a 15psi margin around the desired set point as the tanks drain by using a bang-bang solenoid valve controller and 3000psi nitrogen. The pressurization controller also utilizes the tank vent valve in the case that there is significant overshoot on the tank pressure to limit pressure fluctuations making it to the engine.

From the pressurization data, it is clear that the pressure swing was very significant. This is from two factors. One was the software buffer for the data was not being flushed frequently enough, making the response time of the controller in excess of 150ms. This slow response was compounded by a second factor, the extremely high pressurization flow rates through the press system. As evident in the video of the test, the pressurization swings were large enough to consistently trigger the vent valves. The pressure swings are evident in the graph of the oxidizer test tank pressure below. The dotted lines represent the set points for the controller, and the solid line the goal pressure.


These two issues are very straightforward to fix. By flushing data buffers more frequently, the response time of the system can be brought down significantly. Response time can be further improved by hard coding the controller onto the DAQ and control board, Hyperion. Combining this with orifices on the nitrogen feed to limit flow, the bang-bang controller should be tunable to the point that the pressure swing is within our set points and the vent would never need to open.

A second fix that is needed is to the vent valves themselves. On the oxidizer tank, the valve points down the side of the tank. This orientation allowed the thrust from the valve to actually push up on the oxidizer tank load cell, creating fluctuations in weight data with the valve opening. On the fuel tank, a horizontally mounted vent valve was able to swing the tank, as evident in the video. To remedy this, both valves will have a T fitting attached to the output, so that there will be no net thrust from the vent valve. The tanks will also be constrained in the stand from horizontal movement. An example of the effects the vent valves had on mass measurement of the fuel tank is given to the right.


Because of the fluctuations in the tank mass during the test, the mass flow rate had to be backed out from the starting and ending mass versus the time the test ran. Since this is not ideal, future testing will focus on obtaining more accurate mass flow data.



For proper ignition of the engine, oxidizer and fuel have to enter the combustion chamber at precisely the right time. This is made difficult by the fact that the fuel has to make it through the regenative cooling channels before reaching the injector. To determine the time from valve open command to arrival of each propellant at the injector and valve close command to the end of propellant flow, the valve command was overlaid on the pressure data from multiple tests. Below, an example from an oxidizer flow test is shown. The difference between the valve command (orange) and the pressure (blue) starting to change was taken as the valve timing. For ox and fuel, start up timing was remarkably consistent, with all tests being within 3ms and 12ms of each other, respectively. Shut down timing was more variable, varying as much as 27ms and 40ms, respectively.

valve timing.PNG



For future testing, the GSE system will implement fixes for the issues described above, providing better mass flow data and less fluctuation in feed pressure. The next test will be a cold flow to verify our GSE corrections. The pressure drop across the oxidizer injector will also be roughly verified by water flow testing. Following the completion of these two steps, hot fire testing will begin. Hot fire testing will start with several short duration tests, focusing on measuring start-up transients and ensuring that the engine can be ignited safely and consistently. Testing will then move on to longer tests focusing on measuring steady state performance as well as varying oxidizer to fuel ratios.


Joseph BeaupreComment