A compact and cost effective GRU flow sensor to estimate propellant flow rate and mixture ratio for payload capacity enhancement in Liquid Propellant Rocket Engines

Abstract

This research focusses on developing a single GRU flow sensor to estimate the volumetric flow rate of propellants, Liquid Hydrogen (fuel) and Liquid Oxygen (oxidiser) and to compute Mixture ratio in Liquid Propellant Rocket Engine (LPRE). This single GRU flow sensor replaces a pair of massive Turbine flow meters in LPRE. This enhances the payload capacity (satellite weight) of the launch vehicle. The raw engine data is collected from the ground hot test of LPRE conducted for a duration of 100 s. The significance of this research is to estimate the flow rate of propellants from the functionally dependent pressure parameters such as Combustion chamber pressure ($P_C$), Fuel injection pressure ($P_1$) and Oxidiser injection pressure ($P_2$). The GRU network learns the temporal flow rate dependencies and estimates the fuel and oxidiser flow rate in the three engine operating phases. Mixture Ratio is computed from the GRU estimated flow rate and compared with the actual. Analysis on transient errors in the engine operating phases, estimation performance evaluation with metrices, Root mean square error (RMSE), Mean absolute error (MAE) and R-squared (R²), and performance agreement using Bland Altman approach conducted to assess the estimation effectiveness of GRU flow sensor. An RMSE of 0.3640 and 0.3725 for fuel and oxidiser flow rate respectively and an error less than ±2 % in computed mixture ratio proves that GRU flow sensor estimation is accurate. Additional analysis on weight and cost from literatures show that the hardware model weighs approximately 1.5 kg with a cost benefit of around $71,000. This facilitates a three - fold enhancement in payload capacity of launch vehicle.