The continuous pursuit of enhanced performance, environmental compatibility, and cost efficiency in hybrid rocket engines (HREs) has led to the exploration of nano-additives as energetic performance enhancers. This research investigates the regression behaviour and performance prediction of hybrid rocket engines enhanced with nano-additives using an Adaptive Gaussian Process Regression with Principal Component Reduction (Adaptive GPR-PCR) framework. Seven additives- Aluminium (Al), Boron (B), Sodium Borohydride (NaBH4), Potassium Borohydride (KBH4), Potassium Nitrate (KNO3), Lithium Aluminium Hydride (LiAlH4), and Lithium Borohydride (LiBH4) -were analysed based on their thermal conductivity, decomposition enthalpy, and hydrogen yield to quantify their influence on regression rate (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">$\dot{r}$</EquationSource> <EquationSource Format="MATHML"><math> <mover accent="true"> <mi>r</mi> <mo>˙</mo> </mover> </math></EquationSource> </InlineEquation>), specific impulse (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">$Isp$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi mathvariant="italic">Isp</mi> </mrow> </math></EquationSource> </InlineEquation>), and combustion efficiency (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">$\eta_{c}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>η</mi> <mi>c</mi> </msub> </math></EquationSource> </InlineEquation>). The developed correlations incorporate additive effectiveness, particle size, and oxidizer flux into a multi-parametric regression model validated using simulations. The Adaptive GPR-PCR model demonstrated high sensitivity in predicting parameters variation within limit and accurately captured nonlinear dependencies arising from particle chemistry and oxidizer interaction. A detailed mathematical formulation integrating physicochemical parameters with statistical learning enabled the derivation of hybrid performance maps, providing insight into additive-specific behaviour. Among all additives, Hydride based additives exhibited the good results, attributed to its elevated hydrogen release and moderate decomposition temperature. The framework effectively bridges empirical combustion data with computational learning, enabling a robust predictive methodology for nano-additive hybrid fuels.

Regression analysis on the selection of hybrid rocket engine with suitable nano–additives

期刊论文
发表日期：

2026年1月31日

查看全文

Harikrishna Chavhan,
Akash Pawar,
Amit Kumar Thakur

摘要

The continuous pursuit of enhanced performance, environmental compatibility, and cost efficiency in hybrid rocket engines (HREs) has led to the exploration of nano-additives as energetic performance enhancers. This research investigates the regression behaviour and performance prediction of hybrid rocket engines enhanced with nano-additives using an Adaptive Gaussian Process Regression with Principal Component Reduction (Adaptive GPR-PCR) framework. Seven additives- Aluminium (Al), Boron (B), Sodium Borohydride (NaBH₄), Potassium Borohydride (KBH₄), Potassium Nitrate (KNO₃), Lithium Aluminium Hydride (LiAlH₄), and Lithium Borohydride (LiBH₄) -were analysed based on their thermal conductivity, decomposition enthalpy, and hydrogen yield to quantify their influence on regression rate ( $\dot{r}$ $\dot{r}$ ), specific impulse ( $Isp$ $Isp$ ), and combustion efficiency ( $\eta_{c}$ $η_{c}$ ). The developed correlations incorporate additive effectiveness, particle size, and oxidizer flux into a multi-parametric regression model validated using simulations. The Adaptive GPR-PCR model demonstrated high sensitivity in predicting parameters variation within limit and accurately captured nonlinear dependencies arising from particle chemistry and oxidizer interaction. A detailed mathematical formulation integrating physicochemical parameters with statistical learning enabled the derivation of hybrid performance maps, providing insight into additive-specific behaviour. Among all additives, Hydride based additives exhibited the good results, attributed to its elevated hydrogen release and moderate decomposition temperature. The framework effectively bridges empirical combustion data with computational learning, enabling a robust predictive methodology for nano-additive hybrid fuels.