A new system design tool for a hybrid rocket engine

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Acta Astronautica 223 (2024) 119–133

Contents lists available at ScienceDirect

Acta Astronautica
journal homepage: www.elsevier.com/locate/actaastro

Review article

A new system design tool for a hybrid rocket engine


Elena Quero Granado ∗,1 , Jouke Hijlkema 2 , Jean-Yves Lestrade 2 , Jérôme Anthoine 3
ONERA/DMPE, Université de Toulouse, Toulouse, F-31055, France

ARTICLE INFO ABSTRACT

Keywords: A new system design tool for the simulation of a full hybrid rocket engine is developed in this article. This tool
System design tool enables to simulate the behaviour of the engine at different conditions/configurations in several minutes from a
Hybrid rocket engine desktop computer by keeping a balance between accuracy and computation time. This makes its use especially
1.5-D combustion chamber model
attractive for the pre-design phases of the engine. The main components of the hybrid rocket engine from our
Propulsion modelling
laboratory-characterized by a catalytic injection of the oxidizer-, are modelled and implemented in the tool:
the feed/injection sub-system through 0-D models of a mass flow rate regulator and a catalyst; the combustion
chamber, with a 1.5-D model; and the nozzle through a 1-D model. An iterative method is employed to reach
the convergence of pressure in the combustion chamber between these sub-systems. The corresponding set of
equations is solved by a Newton–Raphson technique. Seven experiments performed on our lab-scale engine
were used to validate the system design tool. In five of the simulations, the relative differences of the main
physical quantities with the experiment were below 30%, being the largest errors found in the fuel regression
rate and mixture ratio. The best agreements with the experiments were retrieved for the cases with the largest
oxidizer mass fluxes (above 230 kg/m2 /s) and mixture ratios closest to stoichiometry, defining thus, the range
of applicability of the system design tool. The results presented in this article were issued of a Ph.D. thesis.

1. Introduction On one hand, the development of numerical means and Computa-


tional Fluid Dynamics (CFD) tools have boosted the development of
A Hybrid Rocket Engine (HRE) is a space propulsion system that combustion chamber models for HREs. This has allowed to decrease
relies on the chemical reaction of two propellants stored at different the number of test campaigns and thus, to reduce the associated costs
states of the matter. The oxidizer, generally stored in liquid or gaseous to characterize this kind of engine. CFD models can determine quite
phase in a tank, is injected – via a catalyst or an injector – into the accurately the local fuel regression rate along the wall (spatial and
combustion chamber, where the fuel is stored in the solid state. The temporal average errors below 15% [3,4]) when considering com-
high temperature of the injected gases generated by a pyrotechnic bustion and flame models with a detailed chemistry, flow turbulence
ignition device or a catalytic decomposition yields the pyrolysis and models and the radiation phenomena in the combustion chamber,
evaporation of the solid fuel, allowing the combustion between both while consuming less economic and material/human resources. Sev-
the fuel and the oxidizer. This process is governed by a self-sustained eral steady and transient 2-D/3-D models concerning the combustion
turbulent diffusion flame located inside the boundary layer. Finally, process in a HRE have been validated and published previously in this
the high temperature products are ejected and accelerated through journal [5–8], providing promising results about the simulation of the
the nozzle, generating, thus, the thrust of the engine. This propulsion main physics in this kind of engine. Some of them also focus on the
technology is characterized by a high specific impulse (up to 360 s in different ways to improve the regression of the solid fuel [9–12], one
vacuum conditions [1,2]) superior to that of solid propulsion, and a of the main variables affecting the engine performance. Nevertheless,
throttability capacity combined with a simple and safe operation of the these models mostly involve steady-state simulations of the combustion
engine while presenting a low manufacturing cost. These features make chamber only, and not of the full operation of the engine. Moreover,
this technology very attractive, turning it into a feasible alternative to the corresponding computations can last for several hours, which can
the more classical bi-liquid or solid space propulsion systems. be too large for the engine pre-design phases.

∗ Corresponding author.
E-mail addresses: elena.quero.granado@hotmail.com (E. Quero Granado), jouke.hijlkema@onera.fr (J. Hijlkema), jean-yves.lestrade@onera.fr
(J.-Y. Lestrade), jerome.anthoine@onera.fr (J. Anthoine).
1
Propulsion modelling engineer, Orbex.
2
Research Scientist, Rocket Propulsion Laboratory Unit, Multi-Physics for Energetics Department.
3
Head of the Multi-Physics for Energetics Department.

https://doi.org/10.1016/j.actaastro.2024.06.031
Received 20 January 2024; Received in revised form 25 May 2024; Accepted 16 June 2024
Available online 24 June 2024
0094-5765/© 2024 The Authors. Published by Elsevier Ltd on behalf of IAA. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Nomenclature Subindex

𝑎𝑑 Adiabatic
English
𝑎𝑚𝑏 Ambient conditions
𝐴 Cross-section area, m2 𝑏 Burnt or during engine burning
𝐴𝑑 Arrhenius preexponential constant, m/s 𝑐 Combustion
𝐶𝑑 Discharge coefficient 𝑐𝑎𝑡 Catalyst
𝑐∗ Characteristic velocity, m/s 𝑐ℎ Combustion chamber
𝑐𝑝 Specific heat coefficient, J/kg/K 𝑒 Core flow
𝐷 Diameter, m 𝑒𝑥𝑝 Experiment
 Diffusion coefficient, m2 /s 𝑓 Solid fuel
𝐸𝑎 Activation energy, J/mol 𝑓 𝑖𝑛𝑎𝑙 Results at the end of the burnt
𝐹 Thrust, N 𝑓𝑙 Flame
𝐺 Mass flux, kg/m2 /s ℎ𝑦𝑏 Hybrid phase
ℎ𝑐𝑜𝑛𝑣 Heat convection coefficient, W/m2 /K 𝑖𝑛𝑖𝑡 Initial conditions
ℎ𝐷 Mass diffusion coefficient, m/s 𝑚 Mean value across boundary layer
ℎ𝑡 Total specific enthalpy, J/kg/K 𝑛𝑜𝑧 Nozzle
𝐿 Length, m 𝑛𝑢𝑚 Numerical simulation
𝐿𝑒 Lewis number 𝑜𝑢𝑡 Exit conditions of the nozzle
𝑀 Mach number 𝑜𝑥 Oxidizer
 Molar mass, kg/mol 𝑝𝑦𝑟 Pyrolysis
𝑚 Mass, kg 𝑟𝑒𝑓 Reference value
𝑚̇ Mass flow rate, kg/s 𝑟𝑒𝑔 Mass flow rate regulator
𝑁𝑠𝑝 Number of species in the flow 𝑠 Fuel surface
𝑃 Pressure, Pa 𝑠𝑒𝑛𝑠𝑜𝑟 Sensor measurement
𝑃𝑟 Prandtl number 𝑠𝑡 Stoichiometric conditions
𝑄̇ Heat flux, W/m2 𝑠𝑡𝑑 Deviation of several sensor measurements
𝑅 Radius, m 𝑡ℎ Nozzle throat
𝑅𝑔 Gas constant, J/kg/mol cat, p Catalyst particles
𝑟 Radial coordinate (reference at the combus-
Superindex
tion chamber axis), m
𝑟̇ Fuel regression rate, m/s 𝑔𝑒𝑜 Geometric
𝑇 Temperature, K 𝑣𝑐 Volume conservation
𝑡 Time, s
Acronyms/Abreviations
𝑢 Axial velocity component, m/s
𝑉 Volume, m3 Chemical Equilibrium Applications: CEA
𝑣 Radial velocity component, m/s Computational Fluid Dynamics: CFD
𝑥 Axial coordinate European Space Propulsion System Simulation: ESPSS
𝑌𝑘 Mass fraction of the species 𝑘 European Space Agency: ESA
𝑦 Radial coordinate (reference at the fuel Gas-Surface Interaction: GSI
surface), m High-Density Poly-Ethylene: HDPE
Hybrid Rocket Engine: HRE
Greek
Hybrid with CATalyst: HYCAT
𝛾 Heat capacity ratio Newton–Raphson: NR
𝛥ℎ𝑝𝑦𝑟 Specific enthalpy of pyrolysis, J/kg Oxidizer-to-Fuel ratio: O/F
𝛿 Boundary layer, m
𝛿∗ Displacement thickness, m
𝜀 Relative error, %
On the other hand, several system design tools have been developed
𝜖 Void fraction
in the literature to study a HRE over the whole test duration [13–15].
𝜂 Efficiency
These ones implement 0-D /1-D models of the main components of the
𝜃 Momentum thickness, m
engine based on mass and energy balances. The components generally
𝜆 Thermal conductivity, W/m/K
modelled in these tools are:
𝜇 Dynamic viscosity, Pa s
𝜌 Density, kg/m3 • The feed sub-system with a pressure-regulated feed operation or
𝛴 Expansion ratio of the nozzle blowdown process.
𝜏 Viscous stress term, N/m2 • The combustion chamber through an empirical law describing the
𝜔̇ 𝑘 Production rate of species 𝑘, kg/m2 /s fuel regression phenomenon.
• The nozzle, characterized by a choked flow model.

These tools make use of iterative or sequential algorithms to solve


the corresponding equations and retrieve the temporal evolution of the

120
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

desired variables. In this context, we find the European Space Propul- Table 1
Design characteristics of the hybrid rocket engine from our laboratory.
sion System Simulation tool (ESPSS), a simulation platform of any space
propulsion system (hybrid, solid or bi-liquid) that was developed for Injection Type Catalytic

the European Space Agency (ESA) and is composed of an ensemble of Fuel block External diameter 95 mm
libraries with models of generic components [16]. Furthermore, similar Length ≤750 mm
tools making use of optimization algorithms have been conceived in Combustion chamber Chamber pressure ≤7.5 MPa
Oxidizer mass flow rate ≤0.5 kg/s
literature [17,18]. These ones enable to determine the set of parameters
(spacecraft mass, nozzle expansion ratio, combustion chamber pressure,
mixture ratio, etc.) that provide the best engine design while achieving
the mission requirements (total spacecraft mass, 𝛥𝑣). distinguished: a monopropellant phase, where the oxidizer is injected
Notwithstanding, the empirical coefficients in the regression rate without any combustion taking place; and then, a hybrid phase, where
laws generally employed in the combustion chamber models from these the fuel pyrolysis and the combustion process begin, producing the rise
applications depend on the Oxidizer-Fuel couple (O-F) and the test of pressure and temperature in the combustion chamber. The test ends
conditions that have been used to characterize them, such as the type when the oxidizer valve is closed, entailing a sharp decrease in pressure
of injection, the size of the engine or the geometry of the fuel block, until the atmospheric value.
among others. Hence, they cannot be generally applied into the system The general design conditions of the HYCAT engine are summarized
design tools to simulate any configuration of a HRE. In this context, it in Table 1. The combustion chamber is made up of several cylindrical
becomes necessary the development of a combustion chamber model sections of 250 mm long that are put together to suit the test require-
that keeps a compromise between the computational cost of the 0-D ments and study the impact of different mixture ratio ranges (close to
models and the accuracy of the CFD simulations. stoichiometry, far from stoichiometry, etc.), with the maximum length
An approximate 2-D model of the combustion chamber (called going up to 750 mm. The length of the individual fuel blocks (250 mm),
1.5-D in this article) is developed to broaden the applicability and together with the 95 mm external diameter (thus, inner diameter of
improve the accuracy of the 0-D empirical models implemented in the chamber shell), were chosen to coincide with the values of the fuel
most of the existing system design tools, while still keeping a low blocks used on the laboratory’s other assemblies and take advantage
computation time in comparison with the 2-D/3-D CFD models. This of the existing tooling. To observe experimentally the influence of the
pseudo 2-D model is based on an integral formulation of the boundary oxidizer mass flux (parameter of prime importance in the sizing of a
layer. The integral models are generally less accurate than 2-D/3-D hybrid engine) on the regression rate, the oxidant flow rate is assumed
models. Nevertheless, their main advantage over the latter ones is to cover a range up to 0.5 kg/s, which corresponds to a mass flux of
the significant saving in computational time because of the removal up to 1000 kg/m2 /s, approximately, for the retrieved cross-sections in
of one spatial dimension from the discretized equations. The integral the tests (i.e.: 25 mm of initial internal diameter of the fuel block),
methods consider the information in both spatial directions (radial and an attractive operating range for the HRE applications. Moreover, the
longitudinal), enabling a more physical treatment of the problem, and combustion chamber is made to withstand pressures of up to 7.5 MPa.
thus, a better accuracy of the results in comparison with 0-D models. The design parameters such as the throat diameter of the nozzle or
The 1.5-D models seem to provide, hence, a satisfactory compromise the inner diameter of the fuel block are chosen for the desired operating
between the desired accuracy and the computational cost. point. The throat of the nozzle is an insert, allowing the diameter of
Integral models have been adopted by several researchers before this one to be selected for each test and set the engine’s operating
[19–21] for the 3-D description of the non-reactive boundary layer pressure. This latter is estimated with Eq. (29). To avoid being sonic
flow to reduce the computational cost. Several studies [22,23] have in the channel of the fuel grain, the initial diameter of this one must
shown that the 1.5-D models are less time-consuming in relation to a be larger than the throat diameter of the nozzle. For the performed
full Navier–Stokes solver while still providing satisfactory results. In the tests, the throat diameter has been generally chosen so that the ratio
context of a HRE and the system design tool that we want to develop, between the initial cross-section of the fuel block and the nozzle throat
this kind of model will allow us to simulate the most important physical is larger than 1.5. The nozzle expansion ratio is then calculated so that
phenomena inside the combustion chamber with computation times the nozzle is adapted on ground, while assuming a heat capacity ratio
adapted to the first design phases of the engine. of 1.22 and a chamber pressure that is estimated with Eq. (29) for
In this article, a 1.5-D combustion chamber model and other simple the envisaged flow rate. The characteristics of these parameters for the
0-D/1-D models of the most important components of a hybrid engine specific tests performed in this article are presented in Section 2.2.
are developed. These models are implemented in a new system design Fig. 1 represents a schematic of the main components of the HYCAT
tool. Eventually, the numerical results obtained with this application engine and the location of the sensors. This engine is constituted by the
are assessed and compared with seven tests performed on the lab-scale following main components:
engine from our laboratory, which will enable to provide the final
conclusions concerning the applicability and validity of the tool. • Oxidizer tank. The hydrogen peroxide in the liquid state is stored
in a tank, where the pressure is controlled and maintained approx-
2. Materials and methods imately constant thanks to an external nitrogen pressurization
system. As a result, a roughly constant oxidizer mass flow rate is
In this section, the materials and methods employed in the devel- supplied to the propulsion system. Here, a pressure gauge sensor
opment and validation of the system design tool are presented. provides the pressure measurement.
• Mass flow rate regulator. This element controls the flow of
2.1. Presentation of the lab-scale hybrid rocket engine oxidizer injected from the tank into the combustion chamber. It is
composed of a diaphragm that reduces the passage section of the
HYCAT (HYbrid with CATalyst) is the lab-scale hybrid rocket engine flow, producing a pressure drop that defines the mass flow rate
from our laboratory that has been employed to validate the system that is measured by a Coriolis flowmeter. Here, another pressure
design tool developed in this article. The ignition of this engine is gauge sensor is used to measure the pressure after the diaphragm
made by the injection of high temperature gas into the combustion section.
chamber, which comes from the chemical decomposition of hydrogen • Catalyst. This element produces the chemical decomposition of
peroxide (H2 O2 ) and water mixture in a catalyst. In this case, the liquid hydrogen peroxide and water mixture into a gaseous mix-
ignition process does not happen immediately, and two phases are ture of di-oxygen and water vapour at high temperature (between

121
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Fig. 1. Schematic of the hybrid rocket engine from our laboratory.

900–1200 K depending on the H2 O2 concentration): 𝑥H2 O2 + Table 2


(1 − 𝑥)H2 O → 𝑥2 O2 + H2 O. This element is composed of a Characteristics of the sensors employed on the hybrid rocket engine from our laboratory.

liquid injection plate and a closed Inconel cylindrical chamber Device Uncertainty Operating range

(𝐷𝑐𝑎𝑡,𝑐ℎ = 50 mm and 𝐿𝑐𝑎𝑡,𝑐ℎ = 80 mm) with a refractory steel Coriolis flowmeter 0.1% 0–1 kg/s
Three-component dynamometer −0.2% 0–2 × 103 N
mesh that stores and retains Al2 O3 −Pt particles. Here, three type-
Piezo-resistive pressure transducers 0.2% 7 × 10−2 –1 × 101 MPa
K thermocouple sensors provide the temperature measurements. Type-k thermocouple ±1 K 273.15–1375 K
The thin thermocouples employed in the tests allows us to obtain
a rapid response of the sensors, with no need to account for
any delay nor adjustments of the retrieved values due to thermal
inertia. while employing the oxidizer mass flow rate and the space-average
• Combustion chamber. It is constituted by a stainless steel cylin- chamber pressure from the test.
drical shell containing the fuel grain. For redundancy reasons, In addition to the ballistic reconstruction method, a space–time
two pressure measurements are performed at the entrance of the average value of the regression rate has been calculated for each test
chamber by two piezo-resistive pressure transducers. through Eq. (1). Its value has been obtained by weighting the fuel block
• Post-combustion chamber. This part of the engine is an ex- at the start and the end of the test, allowing the determination of the
tension of the combustion chamber, where the mixing and the total burnt mass after the experiment and thus, the final space-average
combustion of propellants continues, hence, improving the com- diameter through volume conservation [see Eq. (2)]. This quantity
bustion efficiency. Other two piezo-resistive pressure sensors are can also be estimated by geometrically measuring and averaging the
𝑔𝑒𝑜
employed for redundancy reasons. diameter of the block along the axis at the end of the test (𝐷̄ 𝑐ℎ,𝑓 𝑖𝑛𝑎𝑙
). In
• A convergent-divergent carbon Laval nozzle. The throat diameter this article, (̄⋅) symolizes the space-average value of a given quantity.
is measured with pins (0.05 mm accuracy) before and after each 𝐷̄ 𝑐ℎ,𝑓 𝑖𝑛𝑎𝑙 − 𝐷𝑐ℎ,𝑖𝑛𝑖𝑡
test to determine any erosion. Moreover, the HYCAT engine is in- ⟨𝑟⟩
̇ 𝑡= (1)
2𝑡ℎ𝑦𝑏
stalled on a three-component quartz dynamometer thrust bench, √
where the thrust forces are directly measured. 𝑣𝑐
4𝛥𝑚𝑓
𝐷̄ 𝑐ℎ,𝑓 𝑖𝑛𝑎𝑙 =
2
𝐷𝑐ℎ,𝑖𝑛𝑖𝑡 + (2)
𝜋𝜌𝑓 𝐿𝑐ℎ
The characteristics of the sensors employed in the test bench are
summarized in Table 2. In tests carried out with an oxidant flow rate 2.2. Experiments performed on the test bench
of 100 g/s, the flowmeter was capped at 200 g/s and calibrated below
this value, while being capped at 500 g/s for tests with an oxidant To validate the system design tool developed in this article, a total
flow rate greater than 150 g/s. The flow rate is measured at the of seven tests performed on the HYCAT engine has been employed.
outlet of the tank before entering the catalyst, in the liquid phase. These are the tests 12, 23, 26, 46, 47, 48 and 49. Table 3 contains the
The sampling frequency of the pressure sensors incorporated in the geometric conditions of the fuel grain and nozzle for all of them, whilst
combustion chamber is generally set at 10 kHz. A 2 kHz low-pass Table 4 presents the physical parameters of the experiments in time
analogue filter was used. Each of the pressure sensors in the combustion average, such as the oxidizer mass flow rate, the mixture ratio or the
chamber is also equipped with a 200 Hz high-pass analogue filter on total duration of the test (including both monopropellant and hybrid
a duplicated part of the signal. This analogue filtering provides a pre- phases). All of them employed an axial injection of 87.5% H2 O2 as the
processed unsteady signal free of low-frequency phenomena and the DC oxidizer and High-Density Polyethylene (HDPE) as the solid fuel. The
(Direct Current) component. objective behind the realization of these tests was to analyse the effect
A ballistic reconstruction technique has been used [24] to retrieve of the oxidizer mass flux (𝐺𝑜𝑥 ), the chamber pressure, the fuel port size,
the evolution of the space-average regression rate with time. The and the mixture ratio O/F = 𝑚̇ 𝑜𝑥 ∕𝑚̇ 𝑓 on the computed results. More
method assumes a constant value of the combustion efficiency (𝜂𝑐,𝑐ℎ ) details about the outcome of these firings are presented in Section 4.

122
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Table 3 experiment and quantify the inaccuracy of the measurement technique,


Geometric characteristics of the thrust chamber of our hybrid rocket engine for the four repeated measurements on the same grain were performed at
performed tests.
the same axial locations. The overall mean deviations of the diameter
Test 12 23 26 46 47 48 49
along the axis were also calculated by using the one-sided t -distribution
𝐷𝑐ℎ,𝑖𝑛𝑖𝑡 , mm 25 25 25 25 40 25 40 with 95% confidence interval. The resulting deviations ended up being
𝐿𝑐ℎ , mm 239.9 244.64 240 249 250 498.8 499.2
𝐷𝑡ℎ , mm 7 12.4 12.4 16 16 16 16
below 2.7%, with the largest errors near the beginning and end of the
𝛴 6.3 6.3 6.3 4.3 4.3 4.3 4.3 fuel block.
Eventually, the errors of the rest of the variables that have been de-
rived from the measured quantities through a mathematical expression
Table 4
𝑓 (𝜙1 , 𝜙2 , …) are estimated by subtracting the maximum and minimum
Tests conditions on our hybrid rocket engine (time average values).
values obtained when including the upper and lower error margins of
Test 12 23 26 46 47 48 49
the different measured quantities. Some of the corresponding variables
𝑚̇ 𝑜𝑥 , g/s 97.7 105.1 204.4 302.6 328.5 337.8 348.3 2 ∕4), the regression rate
are the oxidant mass flux 𝐺𝑜𝑥 = 𝑚̇ 𝑜𝑥 ∕(𝜋𝐷𝑐ℎ
𝐺𝑜𝑥 , kg/m2 /s 169.8 144.6 278.6 374.8 187.9 395 230.1
Test duration, s 8.2 20.2 10.0 10.1 10.1 10.1 10.2 [Eq. (1)], the mixture ratio O/F= 𝑚̇ 𝑜𝑥 ∕(𝛥𝑚𝑓 ∕𝑡𝑏 ) or the characteristic
velocity efficiency [see Eq. (29)].

3. Theory and calculation


Test 12 explores the low oxidizer mass flux region of the engine,
whereas test 23, with a very similar oxidizer mass flow rate but, a In this section, the models of the main components of the HYCAT
nozzle of larger throat section was focused on the chamber pressure engine are developed. These ones are: the mass flow rate regulator, the
effect. The duration of the test was also the longest of all, enabling, catalyst, the combustion chamber and the nozzle. The algorithm of the
in this way, to cover a broader range of 𝐺𝑜𝑥 than test 12 and analyse system design tool implementing all of them is presented afterwards.
the impact of the throat erosion over time. Test 26 was performed to
study the impact of a larger mass flux on the regression rate. Finally, 3.1. 0-D mass flow rate regulator model
tests 46, 47, 48 and 49 were performed at an even higher oxidizer mass
flow rate (around 300 g/s) and with different diameter sizes and fuel The oxidizer mass flow rate (𝑚̇ 𝑜𝑥 ) injected into the combustion
grain lengths. The objective was to analyse the impact of the oxidizer chamber through the regulator is mainly influenced by the diaphragm
mass flux, the engine size and the Oxidizer-to-Fuel ratio (O/F) on the section (𝐴𝑑𝑖𝑎𝑝ℎ ) and the pressure difference through the diaphragm
performances. (𝛥𝑃 = 𝑃𝑡𝑎𝑛𝑘 − 𝑃𝑖𝑛𝑗 ). Here, 𝑃𝑡𝑎𝑛𝑘 represents the pressure of the liquid
Moreover, to reduce/avoid the pyrolysis of the lateral face close to oxidizer in the tank prior to the regulator entrance, whereas 𝑃𝑖𝑛𝑗 is
the injector, the distance between this face and the oxidizer injector the pressure at the regulator outlet (equivalent to the pressure at the
ring was diminished in tests 46, 47, 48 and 49. This has also an effect catalyst entrance). In the HYCAT engine, the flow coming from the tank
on the total consumed mass of fuel and, thus, on the estimation of the and passing through this component is still completely in the liquid
time-average regression rate calculated through volume conservation state, and thus, can be treated as incompressible. The estimation of 𝑚̇ 𝑜𝑥
[Eq. (2)]. is, therefore, made through the steady incompressible flow discharge
coefficient equation [Eq. (4)][27]. Here, the variations of density and
2.3. Error estimation temperature are considered negligible and no heat losses through the
walls are contemplated neither.
The technique to quantify the errors in the experimental results is √
presented here. The error of a given quantity 𝜙 that has been obtained 𝑚̇ 𝑖𝑛𝑐 = 𝐴𝑑𝑖𝑎𝑝ℎ 𝐶𝑑 2𝜌(𝑃𝑡𝑎𝑛𝑘 − 𝑃𝑖𝑛𝑗 ) (4)
from a sensor measurement accounts for two components: the accuracy
The coefficient 𝐶𝑑 accounting for the viscous losses through the
of the sensor itself, 𝛿𝜙𝑠𝑒𝑛𝑠𝑜𝑟 , which is estimated with the information in
regulator has been characterized based on previous data from tests
Table 2, and the deviation with respect to the average value between
performed on the HYCAT engine. The corresponding law is given in
the different sensor measurements of this quantity, 𝛿𝜙𝑠𝑡𝑑 . For this latter,
Eq. (5). This curve shows a diminution of 𝐶𝑑 with the increase in the
the one-sided Student’s t -distribution with 95% confidence interval
oxidant mass flow rate (𝑚̇ 𝑜𝑥 ) that is linked to the geometry impact
has
√ been employed. Hence, the total error is estimated as 𝛿𝜙𝑡𝑜𝑡𝑎𝑙 = (𝐴𝑑𝑖𝑎𝑝ℎ ) on 𝑚̇ 𝑜𝑥 . The diameters of the orifices employed in the HYCAT
𝛿𝜙2𝑠𝑒𝑛𝑠𝑜𝑟 + 𝛿𝜙2𝑠𝑡𝑑 . tests have been: 1.4 mm for tests 12 and 23; 1.9 mm for test 26; and
Concerning the estimation of the burn time error 𝛿𝑡𝑏 , the method 2.6 mm for tests 46 to 49, corresponding to the different mass flow
of Durand et al. [25] has been employed. In this one, the peaks of rates tested (100 g/s, 200 g/s and 300 g/s, respectively).
the quadratic module of the chamber pressure derivative | 𝑑𝑃 𝑑𝑡
|2 enable
to identify the start up and shutdown phases of the engine. Then, 𝐶𝑑 = −0.6008𝑚̇ 2𝑜𝑥 − 0.4608𝑚̇ 𝑜𝑥 + 0.8704 (5)
the error in the burn time is retrieved by using the propagation of To simulate the whole operation of the engine (including the tran-
errors technique for the start (𝛿𝑡𝑠𝑡𝑎𝑟𝑡 ) and end phase (𝛿𝑡𝑓 𝑖𝑛𝑎𝑙 ) incertitudes sient regime), a linear rise of 𝑚̇ 𝑜𝑥 during 𝑡𝑡𝑟𝑎𝑛𝑠,𝑟𝑒𝑔 from a zero mass flow
[Eq. (3)]. In Eq. (3), 𝛿𝑡𝑠𝑡𝑎𝑟𝑡 and 𝛿𝑡𝑓 𝑖𝑛𝑎𝑙 are computed as the difference up to the steady value of Eq. (4) has been implemented. Hence, the
between the first occurrence of a maximum peak and a second one, temporal law of the mass flow rate results in: 𝑚̇ 𝑜𝑥 (𝑡) = 𝑚̇ 𝑖𝑛𝑐 𝑡∕𝑡𝑡𝑟𝑎𝑛𝑠,𝑟𝑒𝑔 .
located downwards, at 1% of the previous peak value. Here, 𝑡𝑡𝑟𝑎𝑛𝑠,𝑟𝑒𝑔 was defined to be 0.18 s in agreement with the tests

𝛿𝑡𝑏 = 𝛿𝑡2𝑠𝑡𝑎𝑟𝑡 + 𝛿𝑡2𝑓 𝑖𝑛𝑎𝑙 (3) performed.

For the computation of the error of the fuel block diameter mea- 3.2. 0-D catalyst model
sured along the axis after the burn, the non-destructive method based
on the communicating vessels developed by Glaser et al. [26] is em- In the space propulsion context, several researchers have developed
ployed. This one consists of connecting the fuel grain to a transparent complex 1-D and 2-D CFD models of a reactive two-phase flow in
tube, which is being filled with water until the fluid level reaches a a catalyst by simulating the decomposition reaction kinetics with an
predefined distance 𝑥. Here, the possible errors can come from the Arrhenius law [28,29]. Nevertheless, to ease the weight of the compu-
capillarity effect, parallax of the vessel employed to measure the water tations in the system design tool, a simple 0-D model for the catalytic
height, the syringe filling, etc.. To assess the reproducibility of the injection system has been developed in this research.

123
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

The temperature (𝑇𝑐𝑎𝑡 ), pressure (𝑃𝑐ℎ,0 ), and oxidizer mass flow rate being considered. The density and viscosity in Eq. (8) were computed
(𝑚̇ 𝑜𝑥 ) at the exit of the catalyst are required for the simulation of by using the average temperature and pressure between the inlet and
the flow in the combustion chamber. The temperature (𝑇𝑐𝑎𝑡 ) is esti- the exit of the catalyst.
mated with Gordon and McBride’s code [30] of Chemical Equilibrium ( )
Applications (CEA) as the adiabatic temperature from the chemical 𝐿𝑐𝑎𝑡,𝑐ℎ 1 − 𝜖 𝑢̄ 𝑐𝑎𝑡 𝐺𝑜𝑥 𝑢̄ 𝑐𝑎𝑡
𝛥𝑃 = 𝑃𝑖𝑛𝑗 − 𝑃𝑐𝑎𝑡 = 4826.1(1 − 𝜖)𝜇 + 56.30
decomposition of H2 O2 –H2 O for a given initial temperature (𝑇𝑖𝑛𝑗 ) and 𝑔 𝜖3 2
𝐷𝑐𝑎𝑡,𝑝 𝐷𝑐𝑎𝑡,𝑝
pressure (𝑃𝑖𝑛𝑗 ) of the reactants. The actual (experimental) decomposi-
(8)
tion temperature is, in general, lower than the adiabatic one given by
CEA. This effect is accounted for through a decomposition efficiency Eq. (8) is formulated in the steady regime. Therefore, the transient
(𝜂𝑐,𝑐𝑎𝑡 ) that has been defined between 0.94 and 0.98 for our catalyst. evolution of pressure in the monopropellant phase cannot be correctly
A region of temperature growth during the monopropellant phase simulated with this one. Indeed, at the beginning of this phase, very
from 𝑇𝑖𝑛𝑗 (≈290 K) up to the steady decomposition value of H2 O2 has large 𝛥𝑃 associated with a rapid increase of the oxidizer mass flow
been observed in the experiments. This process of temperature increase rate have been obtained in the simulation, yielding a negative outlet
can last several hundreds of milliseconds in a test of a few seconds. pressure 𝑃𝑐𝑎𝑡 . To avoid this, the instantaneous change in the flow
The transient phase might have, thus, a non-negligible impact on the velocity 𝑑 𝑢̄ 𝑐𝑎𝑡 ∕𝑑𝑡 has been limited to 6–30 m/s2 according to the tests
fuel pyrolysis start. To reproduce this phenomenon in a simple way, the performed.
experimental evolution of the temperature has been approximated by a
linear rise during 𝑡𝑡𝑟𝑎𝑛𝑠,𝑐𝑎𝑡 from the temperature at the catalyst entrance
3.3. 1.5-D combustion chamber model
(𝑇𝑖𝑛𝑗 ) up to the decomposition value defined by 𝑇𝑐𝑎𝑡 = 𝜂𝑐,𝑐𝑎𝑡 𝑇𝑐𝑎𝑡,𝑎𝑑 [see
Eq. (6)].
The physics in the combustion chamber of a HRE have been de-
{ 𝜂𝑐,𝑐𝑎𝑡 𝑇𝑐𝑎𝑡,𝑎𝑑 (𝑌𝐻 𝑂 ,𝑃𝑐𝑎𝑡 ,𝑇𝑖𝑛𝑗 )−𝑇𝑖𝑛𝑗
𝑇𝑖𝑛𝑗 + 𝑡 2
for 𝑡 < 𝑡𝑡𝑟𝑎𝑛𝑠,𝑐𝑎𝑡 scribed with a 1.5-D model, an approximate 2-D model that is based
𝑇𝑐𝑎𝑡 = 𝑡𝑡𝑟𝑎𝑛𝑠,𝑐𝑎𝑡 (6) on the integration of the flow equations in the radial direction and the
𝜂𝑐,𝑐𝑎𝑡 𝑇𝑐𝑎𝑡,𝑎𝑑 (𝑌𝐻2 𝑂 , 𝑃𝑐𝑎𝑡 , 𝑇𝑖𝑛𝑗 ) for 𝑡 ≥ 𝑡𝑡𝑟𝑎𝑛𝑠,𝑐𝑎𝑡
numerical simulation of the resulting equations along the channel axis.
The parameter 𝑡𝑡𝑟𝑎𝑛𝑠,𝑐𝑎𝑡 has been estimated with the experimental The main hypotheses behind the 1.5-D model developed in this article
data from Lestrade et al. [31] and for the tests 12, 23 and 26. In are the following:
particular, it was found that pre-heating the catalyst, decreasing the
size of the catalyst particles or increasing the oxidizer mass flow • The oxidizer is axially injected into the combustion chamber. It is
rate contributed to reduce the duration of this phase. An empirical considered completely in the gaseous phase and described as an
expression to estimate 𝑡𝑡𝑟𝑎𝑛𝑠,𝑐𝑎𝑡 was developed as a function of 𝑚̇ 𝑜𝑥 , ideal gas mixture [Eq. (9)]. No jet impingement nor vortex effects
the dimensions of the catalyst chamber (𝑉𝑐𝑎𝑡,𝑐ℎ and 𝐷𝑐𝑎𝑡,𝑐ℎ ), the void produced by a swirl injection are modelled.
fraction (𝜖), the average diameter (𝐷𝑐𝑎𝑡,𝑝 ) and density of the particles ⎛𝑁 ⎞

𝑒𝑠𝑝
𝑅 ⎟
(𝜌𝑐𝑎𝑡,𝑝 ), and the ratio between a reference temperature 𝑇𝑟𝑒𝑓 and the 𝑃 = 𝜌𝑒 ⎜ 𝑌𝑒 𝑇 (9)
⎜ 𝑘 𝑘 𝑘 ⎟ 𝑒
initial one inside the catalyst 𝑇𝑐𝑎𝑡,𝑖𝑛𝑖𝑡 [Eq. (7)]: ⎝ ⎠
( ) ( ) ( )𝑞
𝜌𝑐𝑎𝑡,𝑝 𝑉𝑐𝑎𝑡,𝑐ℎ (1 − 𝜖) 𝑚 𝐷𝑐𝑎𝑡,𝑝 𝑝 𝑇𝑟𝑒𝑓 • In the developing flow configuration, the flow is divided into
𝑡𝑡𝑟𝑎𝑛𝑠,𝑐𝑎𝑡 = 𝐵 (7)
𝑚̇ 𝑜𝑥 𝐷𝑐𝑎𝑡,𝑐ℎ 𝑇𝑐𝑎𝑡,𝑖𝑛𝑖𝑡 a non-viscous zone (core flow) and a viscous region (boundary
The void fraction 𝜖 that appears in the expression has been obtained layer).
by using an empirical correlation based on an approximation for spher- • 𝑃 𝑟 ≈ 1 and 𝐿𝑒 ≈ 1, meaning the same  for all the species in the
ical catalyst particles: 𝜖 = 0.375 + 0.34𝐷𝑐𝑎𝑡,𝑝 ∕𝐷𝑐𝑎𝑡,𝑐ℎ [32]. The resulting flow.
formula to estimate the time of temperature rise in the catalyst has been • The flame is considered infinitely thin and described by an in-
built for a catalytic chamber of specific dimensions (𝐷𝑐𝑎𝑡,𝑐ℎ = 50 mm finitely fast chemistry (equilibrium) at stoichiometric conditions,
and 𝐿𝑐𝑎𝑡,𝑐ℎ = 80 mm) filled with Al2 O3 −Pt particles assumed of identical where the thermodynamic properties are obtained from Gordon
size. The empirical coefficients in this expression were obtained by a and McBride’s code [30] in a constant temperature–pressure prob-
nonlinear regression fitting (coefficient of determination of 0.898 and lem. Other researchers in the literature have also used chemical
root mean squared error of 0.0483), resulting in: 𝐵 = 11.2589 (s0.3139 equilibrium codes, such as CEA, to describe the combustion pro-
units), 𝑚 = 0.6861, 𝑝 = 0.3689 and 𝑞 = 0.1424. The use of this expression cess and calculate the heat released [34,35]. The corresponding
for other catalytic injection system with different particles, storage models were validated with experimental results and presented a
conditions, or dimensions of the catalytic chamber would require the good agreement with the pressure and regression rate data.
adjustment of these coefficients. The goal of the 1.5-D combustion chamber model is to be used
The energy of the flow and the adiabatic temperature from CEA for the rapid assessment of the performances of a hybrid rocket
are influenced by the pressure loss across the catalyst bed. The Ergun engine (few minutes of computation time on a desktop computer).
equation for packed beds [33] [Eq. (8) in International System units], Therefore, as a first approach, the combustion is described here by
has been applied to this extent. Other researchers have already used a single global reaction: F + 𝑛𝑜 𝑥O𝑥 → 𝑛CO2 CO2 + 𝑛H2 O H2 O, where
this equation to estimate the pressure loss in reacting flows [28,29], only CO2 and H2 O are considered as the main products. The use
obtaining satisfactory results in comparison with the experiments. The of a more detailed chemistry model would certainly increase the
expression can be used for gas or liquid monophase flows, and depends accuracy of the results, but it would also entail a non-negligible
on the unit depth of the packed bed 𝐿𝑐𝑎𝑡,𝑐ℎ , the particle size 𝐷𝑐𝑎𝑡,𝑝 , increase in the computational cost. Despite this, the impact of
the porosity or catalyst void fraction 𝜖, and physical flow parameters using a more accurate combustion model on the results through
such as the density and the average velocity 𝑢̄ 𝑐𝑎𝑡 in the catalyst. The a kinetic reduced mechanism will be investigated further in the
pressure calculated at the exit of the catalyst 𝑃𝑐𝑎𝑡 is considered the future.
same as the pressure at the entrance of the combustion chamber 𝑃𝑐ℎ,0 , Several researchers [36–42] have also described the combustion
since no significant pressure loss from one element to the other was process in a hybrid rocket engine through a single global reaction.
retrieved experimentally. Moreover, on account of the complexity of All of them, employed 2-D equations to describe the gas flow in
the physics involved, the transformation of the fluid from the liquid to the chamber, together with turbulence models (such as SST k-
the gaseous state was not modelled here, with only the gaseous phase omega), while using CFD to solve the corresponding equations.

124
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Their models were successful in predicting the time-average re- the skin friction coefficient and different boundary layer thick-
gression rate along the chamber length in comparison with the ness characteristics (displacement 𝛿 ∗ and momentum thickness 𝜃)
experimental data (errors between 5% and 15%), showing a good proved to be well characterized by the model and showed average
agreement on the chamber pressure values (deviations of less than relative differences with the references below 30%. More details
9% [40]). Moreover, some of these researchers [38,39] observed about this validation can be found in Quero et al.’s [47] work.
the expected behaviour of the regression rate when increasing • The viscous stress term on the fuel surface is represented by a
the oxidizer mass flow rate. However, the assumption of global skin friction coefficient 𝐶𝑓 that is modified with respect to the
chemistry limits the prediction of the spatial heat release rate non-blowing case (𝐶𝑓0 ) to take into account the mass injection
in the flame, which will affect the heat transfer to the fuel and of pyrolysed fuel through the blowing number 𝐵 [Eqs. (12) and
thus, the regression rate [39]. In this regard, Cai and Tian [37] (13)]. 𝐶𝑓0 is estimated by empirical correlations that are defined
found a non-negligible impact on the flame temperature when as a function of 𝑅𝑒 and the geometry along which the flow
using a single global reaction model over a two-step model, and develops [48].
concluded that the two-step chemical reaction model was better 𝐶𝑓
suited than the one-step model to describe the combustion process = 𝐵 −0.68 for 𝐵 > 6.2 (12)
𝐶𝑓 0
in a HRE.
Nevertheless, and, despite the limitations of using a simplified 𝐶𝑓 1
= for 𝐵 ⩽ 6.2 (13)
chemical reaction model for the combustion chamber compo- 𝐶𝑓 0 1 + 0.4𝐵
nent of the system design tool in this research, a good enough
• Only the main gas species from pyrolysis are considered during
estimation of the regression rate and chamber pressure (main pa-
the fuel ablation process: C2 H4 for HDPE, C4 H6 for Hydroxyl-
rameters to determine the engine performances) can be obtained,
Terminated Poly-Butadiene (HTPB), C5 H8 O2 for Poly-Methyl
as it has been observed by other researchers in the literature.
Methacrylate (PMMA), etc. Additionally, no surface reactions
• The flow in the boundary layer is described by empirical radial
other than pyrolysis are considered (oxidation, hydrolysis, etc.).
distributions of the main variables through the Reynolds analogy
• The radiative heat fluxes are neglected in a first instance.
[Eq. (10)], thus closing the system of equations [43]. This analogy
• The curvature effects on the fuel surface are neglected. Consider-
defines an equivalence between the velocity and the temperature
ing the small regression rates in HREs (order of mm/s), especially
and species mass fractions profiles across the boundary layer.
under an axial injection, the shape of the fuel grains does not
In the HRE problem, the presence of the flame will also entail
largely vary along the channel.
a change of sign in the temperature and species mass fraction
gradients. Here, the position of the reaction zone can be retrieved Considering all these hypotheses, the 1.5-D equations defining the
from Eq. (10). core flow in the combustion chamber are defined by Eqs. (14)–(19).
ℎ𝑡 − ℎ𝑡𝑠 𝑌𝑘 − 𝑌𝑘𝑠 [ ]
𝑢 d𝜌𝑒 d𝜌 d𝑢
= = (10) (𝑅 − 𝛿) + 𝑢 𝑒 𝑒 + 𝜌𝑒 𝑒 − 𝜌𝑒 𝑣 𝛿 = 0 (14)
𝑢𝑒 ℎ𝑡𝑒 − ℎ𝑡𝑠 𝑌𝑘𝑒 − 𝑌𝑘𝑠 d𝑡 d𝑥 d𝑥
The empirical, turbulent distribution of the flow velocity across d𝑢𝑒 d𝑢 d𝑃
𝜌𝑒 + 𝜌𝑒 𝑢 𝑒 𝑒 = − (15)
the boundary layer, given in Eq. (11) [44] is employed. In this d𝑡 d𝑥 d𝑥
one, m = 7 for a developing boundary layer, going from 6 to 10 d𝑃
for a fully developed turbulent flow as a function of 𝑅𝑒. =0 (16)
d𝑦
( 𝑦 )1∕𝑚
𝑢 dℎ𝑡𝑒 dℎ𝑡𝑒
= = 𝜂 1∕𝑚 (11) d𝑃
𝑢𝑒 𝛿 𝜌𝑒 + 𝜌𝑒 𝑢 𝑒 = (17)
d𝑡 d𝑥 d𝑡
This allows to account for the turbulence phenomenon in the
d𝑌𝑘𝑒 d𝑌𝑘𝑒
combustion chamber model. To avoid increasing the computation 𝜌𝑒 + 𝜌𝑒 𝑢 𝑒 =0 (18)
d𝑡 d𝑥
time, no complex, turbulence models such as k-epsilon nor SST k-
omega were implemented in our 1.5-D model. Despite the fact 𝑁

𝑠𝑝

that the flow in the combustion chamber of a HRE is highly 𝑌𝑘𝑒 = 1 (19)
turbulent and 3-D [5,6], the simple model approach undertaken 𝑘=0

in this research has enabled, in a first extent, to keep the balance The equations in the boundary layer region are defined in terms
between the computation time and the accuracy of the results, of the displacement thickness 𝛿 ∗ and the momentum thickness 𝜃, both
enabling to complete a simulation of a 10 s firing in around of them linked to the boundary layer thickness 𝛿. These equations are
10 min on a desktop computer. The use of more complex ve- defined by the conservation of mass and the longitudinal quantity of
locity profiles [45], as well as the implementation of a more momentum [see Eqs. (20) and (21)]. The mean density 𝜌𝑚 across the
composite turbulent boundary layer method (i.e.: description of boundary layer has been introduced here to simplify the calculation of
the turbulent shear through an eddy viscosity model [46]) needs the mass flux and quantity of momentum integrals and thus, reduce the
to be further investigated. More complex turbulence models will computational cost.
allow a better representation of the physical phenomena inside d𝛿 d𝜌 d𝜌 d(𝜌 𝑢 ) d𝛿 ∗
the combustion chamber, but they may also produce a significant (𝜌𝑚 −𝜌𝑒 ) +𝛿 𝑚 +(𝑅−𝛿) 𝑒 +(𝑅−𝛿 ∗ ) 𝑒 𝑒 −𝜌𝑒 𝑢𝑒 −𝜌𝑠 𝑣𝑠 = 0 (20)
d𝑡 d𝑡 d𝑡 d𝑥 d𝑥
increase in the computation time, which could compromise the
robustness and application of the system design tool for the d𝜌𝑒 d𝑢 d𝛿 ∗ d(𝛿 ∗ + 𝜃)
(𝑅 − 𝛿 ∗ )𝑢𝑒 + (𝛿 − 𝛿 ∗ )𝜌𝑒 𝑒 − 𝜌𝑒 𝑢𝑒 − 𝜌𝑒 𝑢2𝑒
pre-design phases of the engine. d𝑡 d𝑡 d𝑡 d𝑥
Nevertheless, to validate the capability of this model to reproduce [ ∗
] 2 d𝜌𝑒
+ 𝑅 − (𝛿 + 𝜃) 𝑢𝑒
the several physical phenomena encountered in the combustion d𝑥
chamber of a HRE, different simple flow configurations in a non- d𝑢𝑒 { [ ]} d𝑃
+𝜌𝑒 𝑢𝑒 𝑅 + 𝛿 − 2(𝛿 ∗ + 𝜃) = −𝛿 − 𝜏𝑠 (21)
reactive flow were simulated with the 1.5-D model and compared d𝑥 d𝑥
with the literature data. In particular, the flow regimes (laminar The fuel pyrolysis process is described through the Gas-Surface
and turbulent), the compressibility, the impact of a heated wall, Interaction (GSI) model. This one is based on the physical coupling
and the blowing effect phenomena were studied. In all of them, between the mass flow coming from the thermochemical degradation

125
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

of the solid fuel and the main stream flow. It is based on the mass model is implemented for the nozzle. In the future, either a 1.5-D model
(global and species mass fractions) and energy conservation balances like the one employed for the combustion chamber or a 1-D model that
on the solid/gas interface at quasi-steady equilibrium [Eqs. (22), (23) includes the boundary layer and heat loss effect through the nozzle
and (24), respectively]. Here, the enthalpy of pyrolysis 𝛥ℎ𝑝𝑦𝑟 in Eq. (24) walls could be implemented to describe the flow in this component.
is a property of the fuel. The flow in the nozzle defines the pressure in the combustion
chamber, and will impact thus, the HRE performance. The 1-D nozzle
𝜌𝑠 𝑣 𝑠 = 𝜌𝑓 𝑟̇ (22)
⏟⏟⏟ model employed in the simulations presented in this work is based on
⏟⏟⏟
Mass flux of gases Mass flux of fuel by ablation the transient Euler (non-viscous) equations of continuity, momentum
and energy for a compressible and isentropic flow. The properties of

𝜕𝑌𝑘 the flow are considered to be uniform across a section and vary only
𝜌𝑠+ 𝑣𝑠+ 𝑌𝑘+ − 𝜌𝑠+  along the axial direction. In Appendix, the governing equations of this
⏟⏞⏞⏞⏟⏞⏞⏞⏟ 𝜕𝑦 𝑠+
Mass flux of species 𝑘 by the gaseous flow ⏟⏞⏞⏞⏞⏞⏞⏟⏞⏞⏞⏞⏞⏞⏟ model are presented in terms of the dimensionless velocity, pressure
Diffusion mass flux of species 𝑘 and temperature. The value of these quantities at the entrance of
= 𝜔̇ 𝑘 (𝑇𝑠 ) (23) the nozzle are considered equivalent to the ones in the post-chamber
⏟⏟⏟ (𝑃𝑐ℎ,𝑁+1 , 𝑢𝑒,𝑁+1 ). The same discretization of the equations as in the
Production mass flux of species𝑘
by reaction with the fuel surface 1.5-D combustion chamber is used here (see Section 3.3).
The pressure needed in the combustion chamber to attain either
⌋ ⌋
𝜕𝑇 𝜕𝑇 a subsonic or a supersonic flow in the nozzle is given by the sub-
𝛥ℎ𝑝𝑦𝑟 − 𝜆𝑔 =− 𝜆𝑓 (24)
𝜕𝑦 𝑠+ 𝜕𝑦 𝑠− sonic compressible gas flow equation [Eq. (28)] and the choked flow
⏟⏟⏟
Enthalpy of pyrolysis
⏟⏞⏞⏟⏞⏞⏟ ⏟⏞⏞⏞⏟⏞⏞⏞⏟ condition [Eq. (29)], respectively. In the case of a subsonic flow, an
Convective flux Conductive heat flux on the fuel
additional boundary condition in terms of the static pressure at the exit
The species diffusion term in Eq. (23) and gas convection in Eq. (24) of the nozzle is imposed, this latter being equivalent to the ambient
are quantified by a Newton convection-kind of law [Eqs. (25) and (26), pressure: 𝑃𝑜𝑢𝑡 = 𝑃𝑎𝑚𝑏 . For simplification, a throat erosion model is not
respectively]. implemented here. Finally, the thrust of the engine is estimated by
⌋ ( ) Eq. (30).
𝜕𝑌
𝜌𝑠  𝑘 = 𝜌𝑠+ ℎ𝐷 𝑌𝑘𝑓 𝑙 − 𝑌𝑘𝑠 (25)
𝜕𝑦 𝑠+ 𝑚̇ 𝑜𝑢𝑡 = 𝐴𝑜𝑢𝑡 𝑃𝑐ℎ,𝑁+1
⌋ √
𝜕𝑇 ( ) √
𝜆𝑔 = ℎ𝑐𝑜𝑛𝑣 𝑇𝑓 𝑙 − 𝑇𝑠 (26) √ 2𝛾 
√ ⎡( 𝑃 ) 2
𝛾𝑜𝑢𝑡
(
𝑃𝑜𝑢𝑡
) 𝛾𝑜𝑢𝑡 +1 ⎤
𝛾𝑜𝑢𝑡
𝜕𝑦 𝑠+ ×√
𝑜𝑢𝑡 𝑐ℎ,𝑁+1 ⎢ 𝑜𝑢𝑡
− ⎥ (28)
𝑅𝑔 𝑇𝑐ℎ,𝑁 (𝛾𝑜𝑢𝑡 − 1) ⎢ 𝑃𝑐ℎ,𝑁+1 𝑃𝑐ℎ,𝑁+1 ⎥
Moreover, the heat flux of the solid fuel in Eq. (24) is estimated ⎣ ⎦
through the resolution of the heat transient equation by considering
the fuel as a semi-infinite solid. 𝑃𝑐ℎ,𝑁+1 𝐴𝑡ℎ
𝑚̇ 𝑜𝑢𝑡 = (29)
Eventually, the fuel ablation rate is expressed through an Arrhenius 𝜂𝑐,𝑐ℎ 𝑐 ∗
law [Eq. (27)]. All these features make the developed model reusable

for other HRE configurations independently of the test conditions. In ⎛√
√⎧ ⎞
this case, the estimation of fuel consumption only depends on the ⎜√ ( ) 𝛾+1 ⎡ ( ) 𝛾−1 ⎤⎫ ⎟
⎜√⎪ 2𝛾 2 2 𝛾−1
⎢1 − 𝑃 𝛾 ⎪ 𝑃
⎥⎬ + 𝑜𝑢𝑡 − 𝑃 𝐴
𝑎𝑚𝑏 𝑜𝑢𝑡 ⎟

𝐹 = √⎨ 𝑜𝑢𝑡
aerothermomechanical phenomena of the flow instead of the empirical ⎜ ⎪𝛾 − 1 𝛾 + 1 ⎢ 𝑃𝑐ℎ,𝑁+1 ⎥⎪ 𝑃𝑐ℎ,𝑁+1 𝐴𝑡ℎ ⎟
coefficients, contrary to the combustion chamber models employed in ⎜ ⎩ ⎣ ⎦⎭ ⎟
⎝ ⎠
other system design tools from the literature [13–15].
( ) × 𝑃𝑐ℎ,𝑁+1 𝐴𝑡ℎ (30)
𝐸
𝑟̇ = 𝐴𝑑 exp − 𝑎 (27)
𝑅𝑔 𝑇 𝑠
3.5. System design tool algorithm
A finite difference method is used to discretize the equations of the
model. The combustion chamber is divided into 𝑁 points or nodes The models of the engine components that have been developed in
along the axial direction (𝑥𝑖 , with 𝑖 = 1, 2..., 𝑁), separated by 𝛥𝑥. Section 3 are integrated into the system design tool. The algorithm of
An unconditionally stable, two-level implicit method is used for the this tool allows the easy implementation of the corresponding equations
time derivatives: the backward Euler first-order scheme. This makes of the model and the simulation of the whole or part of the HRE in fast
it possible to take longer time steps without the risk of appearance of computation times on a desktop computer.
instabilities and thus, to reduce the number of iterations. Regarding the To achieve the last requirement, an iterative convergence method
spatial scheme, a first-order backward difference scheme is employed. between the three main sub-systems of the engine is employed to solve
Eventually, the set of equations at each node of the chamber is the full system of equations. These ones are: the feed/injection sub-
solved with a globally convergent Newton–Raphson (NR) algorithm system (oxidizer tank, mass flow rate regulator, catalyst, etc.), the
(10−8 residual). This method combines the quadratic convergence of combustion chamber, and the nozzle. Although this method reduces
the classical NR method for the zero-root finding technique of a system the robustness of the code, it enables a much faster simulation of the
of equations with the backtracking line search strategy, guaranteeing whole HRE in comparison when applying the NR technique to the
thus, the progress towards the final solution at each iteration. whole system of equations, where a simultaneous resolution of these
ones is performed.
3.4. 1-D nozzle model The validation of the tool is made thanks to the tests performed on
the lab-scale engine HYCAT. Here, the oxidizer tank pressure (𝑃𝑡𝑎𝑛𝑘 )
A separate nozzle model from the one of the combustion chamber is considered an input of the system. This parameter can be set to a
is developed in this research. The objective is to enable the modularity constant value, follow a given law over time, or be read from a txt file.
of the system design tool and the possibility to easily implement and To start the simulation, all the input data is defined and the flow in the
use different models for a given component. Considering that the key combustion chamber and nozzle is initialized. Then, at the beginning
component in the modelling of a HRE is the combustion chamber and of each time step, an estimation of the pressure at the entrance of
the right estimation of the fuel regression rate, for simplification and, the combustion chamber (𝑃𝑐ℎ,0 ) is considered. This quantity is used to
in order to save computation time, an unsteady, non-viscous 1-D flow close and thus, solve the system of equations of the elements situated

126
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Table 5
Thermophysical properties of the High-Density Poly-Ethylene (HDPE) [49,50] employed in the computations.
𝐴𝑑 , m/s 𝐸𝑎 , J/mol 𝜌𝑓 , kg/𝑚3 𝑐𝑝,𝑓 , J/kg/K 𝜆𝑓 , W/m/K 𝛥ℎ𝑝𝑦𝑟 , J/kg/K 𝑇𝑣𝑎𝑝 , K 𝑇𝑝𝑦𝑟 , 𝐾
4780 125.604 × 103 960 1850 0.4 2.72 × 106 415 720

upwards the combustion chamber (mass flow rate regulator and cat-
alyst), allowing the computation of the oxidizer mass flow rate, the
pressure at the catalyst entrance and the decomposition temperature
at the exit of this one. With these quantities, the combustion chamber
can be completely simulated by using the sequential scheme described
in Section 3.3. This will allow, afterwards, the calculation of the flow
conditions in the post-chamber, which are considered equivalent to the
conditions at the nozzle entrance.
Subsequently, the nozzle equations are solved. This provides the
pressure 𝑃𝑛𝑜𝑧,0 at the entrance of the nozzle necessary to produce the
supersonic or subsonic flow. When the absolute difference between
the pressure from the combustion chamber equations (𝑃𝑐ℎ,𝑁+1 ) and
the one imposed by the flow through the nozzle (𝑃𝑛𝑜𝑧,0 ) is above a Fig. 2. Firing of test 46 on our lab-scale hybrid rocket engine. (For interpretation of
defined tolerance (10−5 ), the pressure at the entrance of the combustion the references to colour in this figure legend, the reader is referred to the web version
′ of this article.)
chamber, 𝑃𝑐ℎ,0 is updated as 𝑃𝑐ℎ,0 = (𝑃𝑐ℎ,0 − 𝑃𝑐ℎ,𝑁+1 ) + 𝑃𝑛𝑜𝑧,0 . Then,

𝑃𝑐ℎ,0 becomes the new pressure estimate employed in the equations
of the components upwards the combustion chamber. The resolution
of the feeding and injection sub-system’s equations will provide a new The graphs presented in this section compare the smoothed exper-
oxidizer temperature and mass flow rate injected into the combustion imental temporal evolution of a given quantity over time (blue curve)
chamber. These new values will generate new 𝑃𝑐ℎ,𝑁+1 and 𝑃𝑛𝑜𝑧,0 , whose with the one obtained by the system design tool (in black). Moreover,
equivalence will be verified again. This process is then repeated until the estimated error of the experimental values (refer to Section 2.3) is
reaching the convergence criteria for the chamber pressure. Once the also depicted by an orange box in the different plots.
condition is satisfied, the fuel block geometry is updated with the
computed regression rate at each point along the axis of the channel, 4.1.1. Results of the injection sub-system
and the new geometry is used for the next time step. Eventually, the The computed physical quantities of the engine components lo-
simulation finishes at the time indicated in the input data file of the cated upwards the combustion chamber (mass flow rate regulator and
system design tool or when all the fuel at any position of the grain has catalyst) are: the oxidizer mass flow rate (𝑚̇ 𝑜𝑥 ), the pressure at the
entrance of the catalyst (𝑃𝑖𝑛𝑗 ) and the exit temperature (𝑇𝑐𝑎𝑡 ). Here, the
been consumed.
experimental, temporal evolution of the oxidizer pressure in the tank
The computation time to simulate a HRE with the models developed
is used as input of the simulation. This quantity undergoes a slight de-
in this article has been found very dependent on the spatial discretiza-
crease during the transient phase (first 0.5 s), remaining approximately
tion of the combustion chamber (𝛥𝑥), but, especially, on the time
constant throughout the test afterwards.
discretization (𝛥𝑡 in seconds) chosen to simulate the full propulsion
In this test, the hybrid phase starts at 0.47 s. Two conditions are
system (𝛥𝑡−0.829 ). Indeed, the computation time of the implemented
0.794 imposed to simulate the beginning of fuel pyrolysis and the combustion
numerical algorithm follows the power law 𝑁𝑐ℎ , with 𝑁𝑐ℎ being
process in the computation. The first one implies that the surface
the number of discretized nodes along the combustion chamber. By
temperature of the first combustion chamber node must be higher than
using an appropriate time step of 0.1 s, this system design tool is
the fusion temperature of the solid fuel (see Table 5). The second
able to complete a simulation of a 10 s firing in around ten minutes
condition requires a temperature of the injected oxidizer high enough
on a desktop computer, resulting satisfactory for the pre-conception
to entail the reaction between the oxidizer and the pyrolysed fuel
phases of a HRE and/or for the preliminary assessment of the engine
gases. According to previous tests performed on the HYCAT engine
performances prior to a testing camping.
(12, 23 and 26) this temperature has been established at 720 K, value
close to the auto-ignition temperature of ethylene (the main substance
4. Results and discussion produced by the pyrolysis of HDPE), which ranges from 720 to 760 K.
Fig. 3 depicts the experimental and numerical pressure variations
This section addresses the experimental validation of the system through time at the entrance of the catalyst (𝑃𝑖𝑛𝑗 ). A rapid increase
design tool with the models developed in Section 3. To this extent, of 𝑃𝑖𝑛𝑗 at the beginning of the monopropellant phase is observed in
the test 46 performed on our HYCAT engine is analysed in more detail both experimental and numerical evolutions. After this first transient
(Sections 4.1.1 and 4.1.2). A time-average summary of the main phys- phase, the pressure remains relatively constant during a short period
ical quantities obtained from the experiments during the hybrid phase of time, increasing quickly when the hybrid phase begins, and dimin-
and their relative differences with the numerical results is presented ishing slightly afterwards. These effects are correctly simulated by the
afterwards (Section 4.2). Based on the results, the validity region for the computation, with the retrieved pressure values being underestimated
simulation tool is defined. In all the computations, the thermophysical during the hybrid phase by 6.1%.
properties of HDPE in Table 5 have been used. The temporal evolution of the oxidizer mass flow rate obtained
experimentally and numerically is illustrated on Fig. 4. At the start
4.1. Comparison with test 46 of the test, the experimental value increases very quickly, reaching a
maximum point after which, it promptly decreases. Then, during the
The comparison between the experimental and numerical results for hybrid phase, the oxidizer mass flow rate 𝑚̇ 𝑜𝑥 keeps growing slightly
test 46 on the HYCAT engine are presented in this section. This test is due to the pressure decrease outside of the regulator 𝑃𝑖𝑛𝑗 (Fig. 3) for
characterized by one of the highest mass flow rates (300 g/s) from all an approximately constant input pressure 𝑃𝑡𝑎𝑛𝑘 [see Eq. (4)]. The mass
the performed tests in our laboratory. Fig. 2 represents the firing of this flow rate regulator model developed in Section 3.1 allows to correctly
test once the steady state is reached. reproduce the experimental evolution of mass flow rate despite not

127
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Fig. 5. Temporal variation of the temperature at the exit of the catalyst for the
Fig. 3. Temporal variation of the pressure at the entrance of the catalyst for the
experiment and the simulation in test 46. (For interpretation of the references to
experiment and the simulation in test 46. (For interpretation of the references to colour
colour in this figure legend, the reader is referred to the web version of this article.)
in this figure legend, the reader is referred to the web version of this article.)

the simple 0-D model developed in this article enables to simulate the
physics in this component [Eq. (6)], correctly reproducing the temporal
changes of the experiment.

4.1.2. Results in the combustion chamber and nozzle


In this subsection, the main variables related to the flow in the
combustion chamber and the nozzle are analysed and compared to the
ones in the test.
Fig. 6 depicts the temporal variation of the space-average pressure
in the combustion chamber (𝑃̄𝑐ℎ ). Like it happens with 𝑃𝑖𝑛𝑗 and 𝑇𝑐𝑎𝑡 ,
𝑃̄𝑐ℎ increases rapidly during the first milliseconds. Afterwards, when
the energy of the flow transmitted to the solid fuel is high enough, the
ignition of the engine happens, initiating the combustion process and
producing an abrupt rise in pressure. Then, during the first half of the
hybrid phase, the chamber pressure grows slightly in the experiment.
This is because the engine is colder at the start of the test, which
entails a less efficient combustion (low 𝜂𝑐,𝑐ℎ ) as more energy is being
lost through the walls. Later on, less energy is employed to heat the
Fig. 4. Temporal variation of the oxidizer mass flow rate for the experiment and the chamber, and the combustion becomes more efficient (higher 𝜂𝑐,𝑐ℎ ),
simulation in test 46. (For interpretation of the references to colour in this figure
making the pressure rise. Then, the decrease in pressure towards the
legend, the reader is referred to the web version of this article.)
end is associated with the diminution of the total mass flow rate
through the nozzle and a possible throat erosion.
The monopropellant phase and the rapid 𝑃̄𝑐ℎ increase at the be-
reaching the peak retrieved in the experiment. Here, the mass flow ginning of the hybrid phase are correctly simulated by the model.
rates are slightly overestimated throughout the test, producing average Nevertheless, the chamber pressure computed by the tool has a contin-
relative differences of 7.1% in the hybrid phase. uous, slightly, decreasing trend during the whole hybrid phase that is
Fig. 5 depicts the experimental and numerical temporal variations only observed towards the end of the test. This difference in behaviour
of the flow decomposition temperature at the exit of the catalyst. is because the associated, aforementioned physical effects (energy loss,
The curve from the test is obtained by averaging the measurements nozzle erosion) have not been considered in the model. Therefore, the
of the temperature sensors whose values differ less than 25% from pressure in the chamber is only influenced by the changes in the mass
the average. This explains, hence, the non-monotonic temperature flow rate, which are linked, at the same time, to the flow in the nozzle.
increase of the experimental (blue) curve (peaks at 0.6 s and 1 s). Despite this difference, the 𝑃̄𝑐ℎ time-average values are quite similar,
The difference in the measurement of one of the thermocouples with differing only by 9.2%. The incertitude of the error in this case is
respect to the other two sensors (of around 30 K during the hybrid related to both the accuracy of the sensors and the standard deviation
phase), is especially large during the transient phase of the catalyst, (95% confidence interval) of the four different measurements in the
where the temperature increases. This discrepancy could be explained chamber (two sensors located at the chamber entrance and two in the
by the hysteresis phenomenon that these thermocouples can experience post-chamber).
between 523 and 873 K (which can lead to several degrees of error), Fig. 7 illustrates the temporal evolution of the space-average fuel
but it could be also due to the possible metallic contact between one regression rate (𝑟)̇ for this test. In most of the tests performed, the
of the thermocouples with the gas injector. experimental values of the burnt rate have remained approximately
When the combustion process begins and the temperature in the cat- constant all over the engine firing. For this test, however, the experi-
alyst reaches the steady state, the corresponding value is very close to mental evolution shows a continuous decrease, agreeing, thus with the
the experimental one, differing, on average, by 3.3% only. Here again, decreasing trend retrieved in the simulation. The diminution of 𝑟̇ in

128
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Fig. 6. Temporal variation of the space-average pressure in the combustion chamber Fig. 7. Temporal variation of the space-average fuel regression rate for the experiment
for the experiment and the simulation in test 46. (For interpretation of the references and the simulation in test 46. (For interpretation of the references to colour in this
to colour in this figure legend, the reader is referred to the web version of this article.) figure legend, the reader is referred to the web version of this article.)

the computation is related to the decrease in the oxidizer mass flux


𝐺𝑜𝑥 . This latter is associated to the increase of the channel’s cross-
section for an approximately constant oxidizer mass flow rate. The
decrease in 𝐺𝑜𝑥 implies a diminution of the axial velocity of the flow
and a thicker boundary layer. The resulting further location of the
flame from the fuel surface diminishes the heat transfer with this one,
reducing the flame temperature and, eventually, the pyrolysis rate [see
Eq. (27)]. The relationship between 𝐺𝑜𝑥 and 𝑟̇ has been studied before
in literature [45,51,52]. The theoretical analyses and semi-empirical
laws have provided the expression 𝑟̇ = 𝑎𝐺𝑜𝑥 𝑛 (𝑎, 𝑛 > 0), implying thus,

an increase of 𝑟̇ with 𝐺𝑜𝑥 , like the one retrieved numerically.


In the same way as for the chamber pressure, the pyrolysis rate is
overestimated during the first half of the hybrid phase, matching the ex-
perimental curve afterwards. The temporal variations of the regression
rate in the experiment (between +49.8%/−9.4% with respect to the
average value) are smaller than in the simulation (+44.9%/−26.9%),
and the corresponding relative differences between the experiment and
the computation become 13.1% on average. Here, the error box in the
experimental curve has been obtained based on the error margins of the Fig. 8. Computed local diameter of the fuel block at different times of the simulation
other physical quantities that intervene in the ballistic reconstruction in comparison with the experimental one at the end of test 46.
method that is employed to calculate the experimental burnt rate.
Eventually, Fig. 8 depicts the comparison between the local di-
ameter of the fuel block computed by the tool at different times of modifying the skin friction coefficient in the integral boundary layer
the simulation and the final diameter obtained from the experimental equations [58]. Moreover, like in our simulation, these authors also
measurements. The grain diameter retrieved numerically at the end dis- found a more severe burnt rate profile as the port diameter enlarged,
plays a very similar shape to the one in the experiment. Here, the fuel moving the maximum fuel pyrolysis point downstream.
consumption attains its highest value near the entrance of the channel, Despite these differences, the relative error between the experiment
where the skin friction is also the largest. Then, it decreases along the and the computation was about 4.3% on average. The incertitude of
axis, settling at a nearly constant value that increases slightly towards the diameter values is represented by error bars along the different
the end. This shape becomes more pronounced as time passes and more longitudinal stations where several measurements were performed by
fuel is consumed. Other experimental and numerical studies have found the communicating vessel technique (Section 2).
this same local trend [52,53]. The diminution of fuel consumption
along the first half of the block and the maximum fuel pyrolysis at
4.2. Summary of all the performed tests
the entrance of the channel also agree with the trends predicted by
Marxman [54] and found by other authors [55,56]. Nonetheless, other
researchers [3,57] found a concave fuel shape at the chamber entrance The rest of the tests performed on the HYCAT engine have also
when using an axial injection of the oxidizer. This shape was due to been simulated with the system design tool. The temporal evolution
the enhanced fuel consumption produced by the oxidizer jet impacting of the main computed quantities for all the tests were similar to the
on the surface and generating a recirculation flow region. The 1-D ones described in Sections 4.1.1 and 4.1.2 for test 46.
nature of the developed 1.5-D combustion chamber model does not Tables 6 and 7 present the time-average values and uncertainties of
enable the simulation of such 2-D effect, mostly related to the injection the main physical quantities during the hybrid phase for all the tests.
phenomenon. However, its effect could be captured in the model by in- Table 8 provides the relative differences and uncertainties between
troducing a pressure loss factor or through semi-empirical correlations the experimental measurements and the numerical values from the

129
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Table 6
Time-average experimental results in tests 12, 23, 26 and 46 during the hybrid phase.
Test 12 23 26 46
𝑃𝑡𝑎𝑛𝑘 , MPa 6.0 ± 1.2 × 10−2 4.02 ± 8.0 × 10−3 6.39 ± 1.2 × 10−2 7.08 ± 1.1 × 10−2
𝑃𝑖𝑛𝑗 , MPa 3.81 ± 7 × 10−3 1.71 ± 3 × 10−3 3.05 ± 6 × 10−3 4.21 ± 8 × 10−3
𝑇𝑐𝑎𝑡 , K 846.1 ± 34.5 868.5 ± 62 905.5 ± 51 870.8 ± 52
𝑚̇ 𝑜𝑥 , g/s 97.7 ± 0.2 105.1 ± 0.2 204.4 ± 0.4 302.6 ± 0.6
𝐺𝑜𝑥 , kg/m2 /s 169.8 ± 2.5 144.6 ± 2.4 278.6 ± 7.4 374.8 ± 9.7
𝑡ℎ𝑦𝑏 , s 6.3 ± 0.31 17.28 ± 1.65 9.09 ± 0.51 9.40 ± 0.11
𝑃̄𝑐ℎ , MPa 3.6 ± 7.1 × 10−2 1.21 ± 1.4 × 10−2 2.24 ± 2.8 × 10−1 2.08 ± 6.6 × 10−2
𝑔𝑒𝑜
𝐷̄ 𝑐ℎ,𝑓 𝑖𝑛𝑎𝑙
, mm 29.12 ± 0.3 35.86 ± 0.4 36.12 ± 0.7 39.13 ± 0.6
𝐷̄ 𝑐ℎ,𝑓
𝑣𝑐
𝑖𝑛𝑎𝑙
, mm 30.01 ± 0.07 37.02 ± 0.07 37.22 ± 0.07 39.21 ± 0.07
𝑔𝑒𝑜
𝑟̇ , mm/s 0.327 ± 0.04 0.314 ± 0.04 0.612 ± 0.07 0.752 ± 0.04
𝑣𝑐
𝑟̇ , mm/s 0.397 ± 0.03 0.348 ± 0.04 0.672 ± 0.04 0.756 ± 0.01
𝛥𝑚𝑓 , g 50 ± 0.2 135 ± 0.2 135.1 ± 0.2 172.7 ± 0.2
O/F 12.31 ± 0.7 13.45 ± 1.3 13.75 ± 0.8 16.46 ± 0.2
𝜂𝑐,𝑐ℎ 0.88 ± 0.04 0.88 ± 0.04 0.85 ± 0.12 0.94 ± 0.04
𝐹, N 197.7 ± 0.4 – 394.2 ± 0.8 –

Table 7 rest of the physical quantities in both tests are pretty well estimated
Time-average experimental results in tests 47, 48 and 49 during the hybrid phase. by the system design tool (discrepancies with the experiments below
Test 47 48 49 26%). Specifically, the oxidizer mass flow rate, catalyst temperature,
𝑃𝑡𝑎𝑛𝑘 , MPa 7.09 ± 1.4 × 10−2 7.48 ± 1.5 × 10−2 7.43 ± 1.5 × 10−2 pressures, and thrust are characterized by relative differences with the
𝑃𝑖𝑛𝑗 , MPa 3.80 ± 8 × 10−3 3.98 ± 8 × 10−3 3.77 ± 8 × 10−3 experiments below 10%, thus constituting a very good approximation.
𝑇𝑐𝑎𝑡 , K 913.0 ± 34.5 909.1 ± 36.9 907.5 ± 52.7
𝑚̇ 𝑜𝑥 , g/s 328.5 ± 0.7 337.8 ± 0.7 348.3 ± 0.7
In agreement with the literature, the larger ⟨𝐺𝑜𝑥 ⟩𝑡 in test 26 with
𝐺𝑜𝑥 , kg/m2 /s 222.1 ± 4.1 395.0 ± 7.5 230.1 ± 3.0 respect to tests 12 and 23 (278.6 kg/m2 /s) generates higher regression
𝑡ℎ𝑦𝑏 , s 9.28 ± 0.08 9.64 ± 0.06 9.68 ± 0.07 rates in both the experiment and the computation. In this case, the
𝑃̄𝑐ℎ , MPa 1.99 ± 8.6 × 10−3 2.66 ± 7.6 × 10−2 2.41 ± 1.0 × 10−2 simulation provides better estimates for all the variables, reducing the
𝑔𝑒𝑜
𝐷̄ 𝑐ℎ,𝑓 , mm 46.79 ± 0.7 41.0 ± 0.5 47.79 ± 0.4
𝑖𝑛𝑎𝑙 relative differences below 13.5%.
𝐷̄ 𝑐ℎ,𝑓
𝑣𝑐
, mm 47.18 ± 0.07 41.51 ± 0.06 48.10 ± 0.06
𝑔𝑒𝑜
𝑖𝑛𝑎𝑙
The large oxidant mass flow rate (𝑚̇ 𝑜𝑥 ) of test 46 (≈300 g/s) in
𝑟̇ , mm/s 0.366 ± 0.04 0.829 ± 0.03 0.403 ± 0.03 comparison with test 26 (≈200 g/s) for a grain of the same geom-
𝑣𝑐
𝑟̇ , mm/s 0.387 ± 0.01 0.856 ± 0.01 0.419 ± 0.01
etry yields a more important oxidizer mass flux too. Therefore, the
𝛥𝑚𝑓 , g 116.8 ± 0.2 408.9 ± 0.2 266.7 ± 0.2
O/F 26.11 ± 0.3 7.97 ± 0.07 12.63 ± 0.1 corresponding regression rate in the simulation becomes also larger
𝜂𝑐,𝑐ℎ 0.94 ± 0.01 0.86 ± 0.03 0.87 ± 0.01 since, as seen previously, the burnt rate (and, thus, 𝑚̇ 𝑓 ) increases with
𝐹, N 542.6 ± 1.1 737.9 ± 1.5 675.0 ± 1.3 𝑚̇ 𝑜𝑥 . Nonetheless, the increase retrieved in 𝑚̇ 𝑓 with respect to 𝑚̇ 𝑜𝑥 is
smaller, and the mixture ratio grows, moving away from the optimum,
stoichiometric value. Under these conditions, the combustion efficiency
decreases, restraining the increment of the regression rate that would
simulation. The figures in bold indicate the variables that have a non-
have been expected in the experiment at this mass flux. The mixture
negligible error (above 30%) with respect to the experimental value.
ratio has a lesser impact in the code because the flame is considered
The relative difference (𝜀) of any numerical quantity (𝜙𝑛𝑢𝑚 ) with respect
at stoichiometric conditions. In the model, O/F only influences the
to the corresponding experimental/reference value (𝜙𝑒𝑥𝑝 ) has been
𝜙 −𝜙 nozzle flow through the combustion chamber pressure. The larger
calculated as 𝜀 = 𝑛𝑢𝑚𝜙 𝑒𝑥𝑝 × 100. In these tables, no information mixture ratio in test 46 (O/F = 16.46) compared with test 26 (O/F
𝑒𝑥𝑝
concerning the thrust of the engine appears for tests 23 and 46 since = 13.75) produces a larger overestimation of the regression rate too.
a problem encountered before the firing prevented us from performing In spite of this, the relative differences between the numerical and the
these measurements. experimental values in test 46 remain below 14%, still offering a very
In addition, two values of the time-space-average regression rate good accuracy for the pre-design phases of a HRE.
⟨𝑟⟩
̇ 𝑡 are shown: one of them has been computed from the geometric Differently, test 47, which is characterized by a larger port size grain
measurements of the diameter through the communicating vessel tech- (almost the double) than test 26, but by a similar oxidizer mass flow
nique, ⟨𝑟⟩ ̇ 𝑔𝑒𝑜
𝑡 , and the other one is based on the volume conservation rate (≈329 g/s), is, thus, characterized by a smaller ⟨𝐺𝑜𝑥 ⟩𝑡 . Moreover,
applied to the grain at the end of the test, ⟨𝑟⟩ 𝑡 . A larger value of ⟨𝑟⟩
̇ 𝑣𝑐 ̇ 𝑣𝑐
𝑡 the corresponding O/F is the largest from all the performed tests (O/F
𝑔𝑒𝑜
than ⟨𝑟⟩̇ 𝑡 has been generally retrieved in the tests since the former = 26.11), being thus, very far from the stoichiometric value (O∕F𝑠𝑡 =
one accounts for the consumed mass of fuel on the lateral faces. 8.33). Considering the significant impact that the mixture ratio has on
In general, the largest relative differences between the experiment the regression rate in the experiment with respect to the simulation, the
𝑔𝑒𝑜 𝑣𝑐
and the computation have been found for 𝑟̇ , 𝑟̇ , 𝛥𝑚𝑓 and O/F. errors in the estimation of ⟨𝑟⟩ ̇ 𝑡 largely increase for this test (41.1%). The
Particularly, the largest discrepancies in the regression rate ⟨𝑟⟩ ̇ 𝑔𝑒𝑜
𝑡 relative differences between the experiment and the computation are,
between the experiment and the simulation have been found in test on average, of −26.3% for O/F, falling below 10.5% for the thrust, the
12 (43.9%), which is characterized by one of the lowest ⟨𝐺𝑜𝑥 ⟩𝑡 of all pressures and the mass flow rate.
the performed tests (169.8 kg/m2 /s). A better agreement between the The increase in the grain length for the same port diameter and flow
numerical and the experimental ⟨𝑟⟩ ̇ 𝑔𝑒𝑜
𝑡 was retrieved in test 23 (24.52% characteristics at the entrance of the channel produces the increase in
error), described by a flow of similar 𝑚̇ 𝑜𝑥 and slightly smaller ⟨𝐺𝑜𝑥 ⟩𝑡 the total consumed mass of fuel. Under these conditions, the mixture
than test 12. The reason for this was mostly due to the longer duration ratio diminishes, getting closer to the optimum value, and thus, improv-
of test 23. Indeed, in agreement with the literature data, 𝑟̇ has shown ing the combustion process and the burnt rate. This is the case of tests
a strong dependence with 𝐺𝑜𝑥 in the 1.5-D model (see Section 4.1.2). 48 and 49 in comparison with tests 46 and 47, respectively. For test
Hence, as 𝐺𝑜𝑥 decreases with time in the simulation (due to the growth 48, the relative differences of the main quantities with the experiment
of the channel cross-section), 𝑟̇ diminishes as well, entailing a smaller remain below 8%, hence, improving the simulation results of test 46.
average regression rate in test 23 over time, and thus, a smaller The numerical results in test 49 have been better estimated than in test
difference with the experimental value. Except ⟨𝑟⟩ ̇ 𝑔𝑒𝑜
𝑡 in test 12, the 47 as well.

130
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Table 8
Relative differences (in %) between the experiment and the simulation for all the tests during the hybrid phase.
Test 12 23 26 46 47 48 49
𝑃𝑖𝑛𝑗 2.3 ± 0.2 −5.3 ± 0.2 0.6 ± 0.2 −6.1 ± 0.2 2.2 ± 0.2 6.0 ± 0.2 9.9 ± 0.2
𝑇𝑐𝑎𝑡 2.0 ± 4.16 1.45 ± 7.24 0.5 ± 5.7 3.3 ± 6.2 2.0 ± 3.9 2.9 ± 4.2 3.8 ± 6.0
𝑚̇ 𝑜𝑥 −1.1 ± 0.2 −2.3 ± 0.2 0.4 ± 0.1 7.1 ± 0.2 −0.3 ± 0.2 −2.6 ± 0.2 −5.3 ± 0.2
𝑡ℎ𝑦𝑏 3.9 ± 5.2 9.4 ± 5.5 −0.5 ± 5.7 −0.5 ± 1.1 1.1 ± 0.8 −2.4 ± 0.6 −1.8 ± 0.8
𝑃̄𝑐ℎ 4.7 ± 2.1 2.6 ± 1.2 2.1 ± 12.8 9.2 ± 3.5 7.4 ± 0.5 −2.3 ± 2.8 2.7 ± 0.4
𝐷̄ 𝑔𝑒𝑜
𝑐ℎ,𝑓 𝑖𝑛𝑎𝑙
7.0 ± 1.1 8.2 ± 1.2 2.8 ± 2.0 4.3 ± 1.6 6.0 ± 1.5 1.1 ± 1.3 4.3 ± 0.9
𝐷̄ 𝑐ℎ,𝑓
𝑣𝑐
𝑖𝑛𝑎𝑙
4.0 ± 0.1 4.8 ± 0.1 0.3 ± 0.2 4.1 ± 0.2 5.1 ± 0.2 0.1 ± 0.2 3.6 ± 0.1
𝑔𝑒𝑜
𝑟̇ 𝟒𝟑.𝟗 ± 𝟏𝟕.𝟔 24.5 ± 16.7 13.4 ± 11.5 13.1 ± 6.3 𝟒𝟏.𝟏 ± 𝟏𝟔.𝟏 5.4 ± 4.4 28.7 ± 8.9
𝑣𝑐
𝑟̇ 17.3 ± 8.6 11.9 ± 11.7 6.7 ± 0.5 12.1 ± 2.2 𝟑𝟏.𝟖 ± 𝟑.𝟑 2.0 ± 1.4 23.2 ± 2.7
𝛥𝑚𝑓 25.6 ± 0.5 17.8 ± 0.2 0.8 ± 0.2 13.1 ± 0.1 𝟑𝟗.𝟎 ± 𝟎.𝟐 0.7 ± 0.1 24.5 ± 0.1
O/F −17.0 ± 4.6 −12.7 ± 8.7 −6.0 ± 0.3 −5.2 ± 1.4 −26.3 ± 1.6 −4.4 ± 0.9 −25.0 ± 0.8
𝐹 9.3 ± 0.2 – −5.8 ± 0.2 – 10.4 ± 0.2 1.7 ± 0.2 5.7 ± 0.2

The main physical phenomena retrieved in a hybrid rocket engine


over time (monopropellant and hybrid phases) have been pretty well
simulated by the tool. However, some temporal trends from the com-
putation did not exactly reproduce the experiment due to the lack of
modelling of certain physics. Such was the case of the combustion
chamber pressure, where neither the heat losses through the walls nor
the combustion efficiency change over time were considered.
The increase in the fuel block length led to the increase in the total
consumed mass of fuel, yielding, for a given oxidizer mass flow rate,
a decrease in the mixture ratio, making it closer to the optimum value
(stoichiometry) and producing, hence, a general rise in the regression
rate associated with the better combustion. Under these circumstances,
the relative differences between the numerical and the experimental
regression rates decreased, as the mixture ratio became closer to the
Fig. 9. Dependence of the space–time-average regression rate with the oxidizer mass stoichiometric value that was employed to describe the flame and the
flux for the performed tests and the numerical results. (For interpretation of the combustion process in the 1.5-D model.
references to colour in this figure legend, the reader is referred to the web version The comparison between the time-average values of the main physi-
of this article.)
cal quantities from the simulations and the tests has shown the greatest
differences in the fuel regression rate, the total consumed mass of fuel
and the mixture ratio. The burnt rate and thus, the fuel consumption
Fig. 9 illustrates the dependence of the space–time-average regres- in the computations have been overestimated, respectively, by around
sion rate with the oxidizer mass flux from the tests (triangle markers) 22.7% and 17.3% on average in all the seven tests. This produced an
and the simulations (square markers). The relationship defines a power underestimation of the mixture ratio (13.8% on average) for a well
law (coefficient of determination of 0.97) showing an increase in the estimated oxidizer mass flow rate (2.8% error). The relative differences
fuel consumption with the oxidizer mass flux 𝐺𝑜𝑥 that is in agreement of the computed regression rate with the experiment can be considered
with several literature data for this O-F couple [59,60]. The maximum satisfactory in five of the seven tests analysed (tests 23, 26, 46, 48 and
relative differences with the references go up to 34%. The curve from 49), where the errors have been below 30%. In particular, the best
the experiment also presents a lesser ⟨𝑟⟩
̇ 𝑡 dependence for low ⟨𝐺𝑜𝑥 ⟩𝑡 (< agreements have been found for the firings of 10 s duration or longer,
200 kg/m2 /s). This is due to the more important radiative heat fluxes characterized by flows of high oxidizer mass fluxes (> 230 kg/m2 /s)
at low 𝐺𝑜𝑥 [52], which contributes to increase the regression rate. and mixture ratios situated the closest to stoichiometry, with equiva-
lence ratios (ratio between the actual and stoichiometric mixture ratio)
5. Conclusions above 0.65.
In conclusion, the system design tool that has been developed in this
This article has addressed the development of a system design tool research has allowed the simulation of the main physical phenomena
able to simulate a full hybrid rocket engine in fast computation times of a hybrid propulsion system, providing, on average, a good accuracy
on a desktop computer. The goal is to use this tool during the first of the main quantities characterizing the flow in this kind of engine.
pre-design phases of the engine. For this purpose, simple models that Therefore, this tool could be employed for the pre-design phases of a
describe the physics of the main components of a hybrid rocket engine hybrid rocket engine under the operating conditions assessed in this
have been developed while considering an axial and catalytic injection article.
of the oxidizer into the combustion chamber. In particular, 0-D models
describing the flow in a mass flow rate regulator and a catalyst; a 1.5-D CRediT authorship contribution statement
model for the combustion chamber; and a 1-D model for the nozzle.
The algorithm implemented in the system design tool to solve the Elena Quero Granado: Writing – review & editing, Writing – orig-
whole set of equations is based on an iterative convergence technique inal draft, Visualization, Validation, Software, Project administration,
between these three main sub-systems of the engine. The numerical Methodology, Investigation, Formal analysis, Data curation, Concep-
method has been developed to enable the simulation of either the whole tualization. Jouke Hijlkema: Writing – review & editing, Validation,
or an ensemble of components thereof. To validate the tool, seven Supervision, Software, Resources, Project administration, Methodology,
experiments performed on the hybrid engine of our laboratory, which Funding acquisition. Jean-Yves Lestrade: Writing – review & editing,
uses hydrogen peroxide as oxidizer and high density poly-ethylene as Validation, Supervision, Resources, Project administration, Methodol-
fuel, have been employed. The retrieved results have shown a good ogy, Data curation. Jérôme Anthoine: Writing – review & editing,
agreement in the estimation of the mass flow rate, pressure and thrust Validation, Supervision, Resources, Project administration, Funding ac-
for all the tests, with time-average differences below 11%. quisition, Conceptualization.

131
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

Declaration of competing interest [11] T. Wei, G. Cai, H. Tian, X. Jiang, Experiment and numerical research on regres-
sion rate of hybrid rocket motor with single-port Wagon wheel fuel grain, Acta
Astronaut. 207 (2023) 265–282, http://dx.doi.org/10.1016/j.actaastro.2023.03.
The authors declare that they have no known competing finan-
021, 0094-5765.
cial interests or personal relationships that could have appeared to
[12] H. Nagata, M. Ito, T. Maeda, M. Watanabe, T. Uematsu, T. Totani, I. Kudo,
influence the work reported in this paper. Development of CAMUI hybrid rocket to create a market for small rocket
experiments, Acta Astronaut. 59 (1–5) (2006) 253–258, http://dx.doi.org/10.
Acknowledgements 1016/j.actaastro.2006.02.031, 0094-5765.
[13] T. Chelaru, F. Mingireanu, Hybrid rocket engine, theoretical model and experi-
ment, Acta Astronaut. 68 (11) (2011) 1891–1902, http://dx.doi.org/10.1016/j.
Special thanks to ONERA, France and CNES, France (French Space
actaastro.2010.12.008.
Agency) for financially supporting this research, as well as to the ON- [14] M. Gieras, A. Gorgeri, Numerical modelling of the hybrid rocket engine perfor-
ERA’s Rocket Propulsion Laboratory for performing the experimental mance, Propuls. Power Res. 10 (1) (2021) 15–22, http://dx.doi.org/10.1016/j.
campaign on the HYCAT engine. jppr.2021.03.001.
[15] P. Zolla, M. Migliorino, D. Bianchi, F. Nasuti, R. Pellegrini, E. Cavallini, A
computational tool for the design of hybrid rockets, Aerotec. Missili Spazio 100
Appendix. Transient 1-D nozzle equations
(4) (2021) 387–397, http://dx.doi.org/10.1007/s42496-021-00085-3.
[16] European Space Propulsion System Simulation, ESPSS, Empresarios Agrupados,
The 1-D transient Euler equations of continuity [Eq. (A.1)], mo- 2013.
mentum [Eq. (A.2)] and energy [Eq. (A.3)] describing the flow inside [17] F. Mechentel, Preliminary Design of a Hybrid Motor for Small-Satellite Propul-
the nozzle of the engine in the system design tool are presented in sion (Ph.D. thesis), Dept. of Aeronautics and Astronautics, Stanford University,
this appendix in terms of the dimensionless quantities: 𝜌′ = 𝜌 𝜌 ; Stanford, CA, 2019.
𝑐ℎ,𝑁+1 [18] L. Casalino, D. Pastrone, Oxidizer control and optimal design of hybrid rockets
𝑇 𝐴 𝑡 𝑃 𝑢
𝑇′ = 𝑇𝑐ℎ,𝑁+1
; 𝐴′ = 𝐴𝑡ℎ
; 𝑡′ = 𝐿∕𝑎𝑐ℎ,𝑁+1
; 𝑃′ = 𝑃𝑐ℎ,𝑁+1
; 𝑢′ = 𝑎𝑐ℎ,𝑁+1
and for small satellites, J. Propuls. Power 21 (2) (2005) 230–238, http://dx.doi.org/
𝑥
𝑥′ = 𝐿𝑛𝑜𝑧
. 10.2514/1.6556.
[19] W.M. Milewski, Three-Dimensional Viscous Flow Computations using the Integral
Boundary Layer Equations Simultaneously Coupled with a Low Order Panel
𝜕 ′ ′ 𝜕 ( )
(𝜌 𝐴 ) + ′ 𝜌′ 𝐴′ 𝑢′ = 0 (A.1) Method (Ph.D. thesis), Dept. of Ocean Engineering, Massachusetts Inst. of
𝜕𝑡′ 𝜕𝑥 Technology, 1997.
( ) [20] B. Mughal, Integral Method for Three-Dimensional Boundary-Layers (Ph.D.
𝜕 ′ ′ ′ 𝜕 ′ ′ ′2 1 ′ ′ 1 𝜕𝐴′

(𝜌 𝐴 𝑢 ) + ′
𝜌 𝐴 𝑢 + 𝑃 𝐴 = 𝑃′ ′ (A.2) thesis), Dept. of Aeronautics and Astronautics, Massachusetts Inst. of Technology,
𝜕𝑡 𝜕𝑥 𝛾 𝛾 𝜕𝑥 1998.
[21] P. Trontin, A. Kontogiannis, G. Blanchard, P. Villedieu, Description and assess-
[ ( ′ ) ]
𝜕 ′ 𝑇 𝛾 ′2 ment of the new onera 2d icing suite igloo2d, in: 9th AIAA Atmospheric and
𝜌 + 𝑢 𝐴′ Space Environments Conference, AIAA Paper 2017-3417, Denver, CO, 2017.
𝜕𝑡′ 𝛾 −1 2
[ ( ′ ) ] [22] C. Bayeux, E. Radenac, P. Villedieu, Theory and validation of a 2D finite volume
𝜕 𝑇 𝛾
+ ′ 𝜌′ + 𝑢′2 𝐴′ 𝑢′ + +𝑃 ′ 𝐴′ 𝑢′ = 0 (A.3) integral boundary-layer method for icing applications, AIAA J. 57 (3) (2019)
𝜕𝑥 𝛾 −1 2 1092–1112, http://dx.doi.org/10.2514/1.J057461.
[23] E. Radenac, A. Kontogiannis, C. Bayeux, P. Villedieu, An extended rough-
References wall model for an integral boundary layer model intended for ice accretion
calculations, in: 2018 Atmospheric and Space Environments Conference, 2018.
[1] M.J. Chiaverini, K.K. Kuo, Fundamentals of Hybrid Rocket Combustion and [24] J. Messineo, K. Kitagawa, C. Carmicino, T. Shimada, C. Paravan, Reconstructed
Propulsion, in: Progress in Astronautics and Aeronautics, vol. 218, AIAA, Reston, ballistic data versus wax regression-rate intrusive measurement in a hybrid
2007. rocket, J. Spacecr. Rockets 57 (6) (2020) 1295–1308, http://dx.doi.org/10.2514/
[2] C. Schmierer, M. Kobald, U. Fischer, K. Tomilin, A. Petrarolo, F. Hertel, 1.A34695.
Advancing europe’s hybrid rocket engine technology with paraffin and LOX, in: [25] J.E. Durand, J.Y. Lestrade, J. Anthoine, Restitution methodology for space and
8th European Conference for Aeronautics and Aerospace Sciences, EUCASS, 2019, time dependent solid-fuel port diameter evolution in hybrid rocket engines,
http://dx.doi.org/10.13009/EUCASS2019-682. Aerosp. Sci. Technol. 110 (2021) 1270–9638, http://dx.doi.org/10.1016/j.ast.
[3] D. Bianchi, G. Leccese, F. Nasuti, M. Onofri, C. Carmicino, Modeling of high 2021.106497.
density polyethylene regression rate in the simulation of hybrid rocket flowfields, [26] C. Glaser, R. Gelain, A. Bertoldi, J. Hijlkema, P. Hendrick, J. Anthoine, Experi-
Aerospace 6 (88) (2019) http://dx.doi.org/10.3390/aerospace6080088. mental investigation of stepped fuel grain geometries in hybrid rocket engines,
[4] G.D. Martino, C. Carmicino, R. Savino, Transient computational thermofluid- in: 9th European Conference for Aeronautics and Space Sciences, EUCASS, Lille,
dynamic simulation of hybrid rocket internal ballistics, J. Propuls. Power 33 France, 2022, http://dx.doi.org/10.13009/EUCASS2022-4448.
(2017) 1395–1409, http://dx.doi.org/10.2514/1.B36425. [27] S.A. Whitmore, S.N. Chandler, Engineering model for self-pressurizing saturated-
[5] A.G. Kushnirenko, L.I. Stamov, V.V. Tyurenkova, M.N. Smirnova, E.V. N2O-propellant feed systems, J. Propuls. Power 6 (4) (2010) http://dx.doi.org/
Mikhalchenko, Three-dimensional numerical modeling of a rocket engine with 10.2514/1.47131.
solid fuel, Acta Astronaut. 181 (2021) 544–551, http://dx.doi.org/10.1016/j. [28] A. Pasini, L. Torre, L. Romeo, A. Cervone, L. d’Agostino, Reduced-order model
actaastro.2021.01.028, 0094-5765. for H2O2 catalytic reactor performance analysis, J. Propuls. Power 26 (3) (2010)
[6] V.B. Betelin, A.G. Kushnirenko, N.N. Smirnov, V.F. Nikitin, V.V. Tyurenkova, L.I. 446–453, http://dx.doi.org/10.2514/1.44355.
Stamov, Numerical investigations of hybrid rocket engines, Acta Astronaut. 144 [29] S.A. Whitmore, C.J. Martinez, D.P. Merkley, Catalyst development for an arc-
(2018) 363–370, http://dx.doi.org/10.1016/j.actaastro.2018.01.009, 0094-5765. ignited hydrogen peroxide/ABS hybrid rocket system, Aeronaut. Aerosp. Open
[7] C. Glaser, J. Hijlkema, J.-Y. Lestrade, J. Anthoine, Influences of steps in Access J. 2 (6) (2018) 356–388, http://dx.doi.org/10.15406/aaoaj.2018.02.
hybrid rocket engines:Simulation and validation on simplified geometries, Acta 00069.
Astronaut. 208 (2023) 1–14, http://dx.doi.org/10.1016/j.actaastro.2023.03.037, [30] S. Gordon, B.J. McBride, Computer Program for Calculation of Complex Chem-
0094-5765. ical Equilibrium Compositions and Applications, Vol. 1311, NASA Reference
[8] X. Meng, H. Tian, X. Chen, X. Jiang, P. Wang, T. Wei, G. Cai, Numerical Publication, 1994.
simulation of combustion surface regression based on Butterworth filter in hybrid [31] J.Y. Lestrade, P. Prévot, J. Messineo, J. Anthoine, S. Casu, B. Geiger, De-
rocket motor, Acta Astronaut. 202 (2023) 400–410, http://dx.doi.org/10.1016/ velopment of a catalyst for highly concentrated hydrogen peroxide, in: Space
j.actaastro.2022.11.003, 0094-5765. Propulsion 2016, Rome, Italy, 2016.
[9] K. Veale, S. Adali, J. Pitot, M. Brooks, A review of the performance and [32] O. Bey, G. Eigenberger, Fluid flow through catalyst filled tubes, Chem. Eng. Sci.
structural considerations of paraffin wax hybrid rocket fuels with additives, Acta 52 (8) (1997) 1365–1376, http://dx.doi.org/10.1016/S0009-2509(96)00509-X.
Astronaut. 141 (2017) 196–208, http://dx.doi.org/10.1016/j.actaastro.2017.10. [33] S. Ergun, Fluid flow through packed columns, J. Chem. Eng. Prog. 48 (2) (1952)
012, 0094-5765. 89–94.
[10] C. Glaser, R. Gelain, A.E.M. Bertoldi, Q. Levard, J. Hijlkema, J.-Y. Lestrade, [34] F. Barato, M. Lazzarin, N. Bellomo, M. Faenza, A. Bettella, D. Pavarin, A
P. Hendrick, J. Anthoine, Experimental regression rate profiles of stepped fuel numerical model to analyze transient behavior and instabilities on hybrid rocket
grains in hybrid rocket engines, Acta Astronaut. 204 (2023) 186–198, http: motors, in: 47th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit,
//dx.doi.org/10.1016/j.actaastro.2022.12.045, 0094-5765. AIAA Paper 2011-5538, San Diego, CA, 2011.

132
E. Quero Granado et al. Acta Astronautica 223 (2024) 119–133

[35] N. Serin, G. Yalcin, A fast computer code for hybrid motor design, eulec, and [49] N. Gascoin, P. Gillard, A. Mangeot, A. Navarro-Rodriguez, Literature survey for
results obtained for HTPB/O2 combination, in: 39th AIAA/SME/SAE/ASEE Joint a first choice of a fuel-oxidiser couple for hybrid propulsion based on kinetic
Propulsion Conference and Exhibit, AIAA Paper 2003-4747, Huntsville, AL, 2003. justifications, J. Anal. Appl. Pyrolysis 94 (2012) 1–9, http://dx.doi.org/10.1016/
[36] A. Antoniou, K. Akyuzlu, A physics based comprehensive mathematical model j.jaap.2011.11.006.
to predict motor performance in hybrid rocket propulsion systems, in: 41st [50] G. Lengellé, B. Fourest, J. Godon, C. Guin, Condensed-phase behavior and
AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, AIAA Paper ablation rate of fuels for hybrid propulsion, in: 29th Joint Propulsion Conference
2005-3541, Tucson, AZ, 2005. and Exhibit, AIAA Paper 93-2413, 1993, http://dx.doi.org/10.2514/6.1993-
[37] G. Cai, H. Tian, Numerical simulation of the operation process of a hybrid rocket 2413.
motor, in: 42nd AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, [51] C. Carmicino, A. Sorge, Role of injection in hybrid rockets regression rate
AIAA Paper 2006-4506, Sacramento, CA, 2006. behaviour, J. Propuls. Power 21 (4) (2005) 606–612, http://dx.doi.org/10.2514/
[38] M. Lazzarin, F. Barato, D. Pavarin, A. Bettella, CFD simulation of regression rate 1.9945.
in hybrid rockets, J. Propuls. Power 29 (6) (2013) 1445–1452. [52] M. Chiaverini, G. Harting, K.K. Kuo, A. Peretz, Regression rate and heat transfer
[39] C.P. Kumar, A. Kumar, Effect of diaphragms on regression rate in hybrid rocket correlations for hybrid rocket combustion, J. Propuls. Power 17 (1) (2001)
motors, J. Propuls. Power 29 (3) (2013) 559–572. 99–110, http://dx.doi.org/10.2514/2.5714.
[40] B. Zhao, N. Yu, Y. Liu, P. Zeng, J. Wang, Unsteady simulation and experimental [53] G. Gariani, F. Maggi, L. Galfetti, Numerical simulation of HTPB combustion
study of hydrogen peroxide throttleable catalyst hybrid rocket motor, Aerosp. in a 2D hybrid slab combustor, Acta Astronaut. 69 (2011) 289–296, http:
Sci. Technol. 76 (2018). //dx.doi.org/10.1016/j.actaastro.2011.03.015.
[41] D. Helman, M. Wolfshtein, Y. Manheimer-Timnat, Theoretical investigation of [54] G.A. Marxman, Combustion in the turbulent boundary layer on a vaporizing
hybrid rocket combustion by numerical methods, J. Combust. Flame 22 (2) surface, Symp. (Int.) Combust. 10 (1) (1965) 1337–1349, http://dx.doi.org/10.
(1974) 171–190. 1016/S0082-0784(65)80268-5.
[42] G. Gallo, S. Mungiguerra, New entrainment model for modelling the regression [55] P. Estey, D. Altman, J. McFarlane, An evaluation of scaling effects for hybrid
rate in hybrid rocket engines, J. Propuls. Power 37 (6) (2021) 893–909. rocket motors, in: 27th Joint Propulsion Conference, AIAA Paper 1991-2517,
[43] J.A. Schetz, Boundary Layer Analysis, Prentice Hall, New Jersey, 1993. 1991, http://dx.doi.org/10.2514/6.1991-2517.
[44] W. Zhi-qing, Study on correction coefficients of laminar and turbulent entrance [56] A. Karabeyoglu, G. Zilliac, B. Cantwell, Development of scalable spacetime
region effect in round pipe, Appl. Math. Mech. 3 (1982) 433–446. averaged regression rate expressions for hybrid rockets, J. Propuls. Power 23
[45] G.A. Marxman, M. Gilbert, Turbulent boundary layer combustion in the hybrid (4) (2007) 737–747, http://dx.doi.org/10.2514/1.19226.
rocket, Symp. (Int.) Combust. 9 (1) (1963) 371–383, http://dx.doi.org/10.1016/ [57] C. Carmicino, A. Sorge, Influence of a conical axial injector on hybrid rocket
S0082-0784(63)80046-6. performance, J. Propuls. Power 22 (5) (2006) 984–995, http://dx.doi.org/10.
[46] H.L. Moses, in: S.J. Kline, M.V. Morkovin, G. Sovran, D.J. Cockrell (Eds.), 2514/1.19528.
A Strip-Integral Method for Predicting the Behavior of Turbulent Boundary [58] K.M. Krall, E.M. Sparrow, Turbulent heat transfer in the separated, reattached,
Layers in Computation of Turbulent Boundary Layers-1968 AFOSR-IFP Standford and redevelopment regions of a circular tube, J. Heat Transfer 88 (1) (1966)
Conference, Stanford University Press, Stanford, CA, 1969. 131–136.
[47] E.Q. Granado, J. Hijlkema, J. Lestrade, J. Anthoine, Pseudo-2-dimensional [59] E. Wernimont, S. Heister, Combustion experiments in hydrogen peroxide/
modeling and validation of a hybrid rocket combustion chamber, J. Propuls. polyethylene hybrid rocket with catalytic ignition, J. Propuls. Power 16 (2000)
Power (2022) 1–16, http://dx.doi.org/10.2514/1.B38682. 318–326, http://dx.doi.org/10.2514/2.5571.
[48] F. Schultz-Grunow, New Frictional Resistance Law for Smooth Plates, Tech. Rep., [60] B. Ahn, H. Kang, E. Lee, Y. Yun, S. Kwon, Design of multiport grain with
National Advisory Comittee for Aeronautics, Washington, 1941. hydrogen peroxide hybrid rocket, J. Propuls. Power 34 (5) (2018) 1189–1197,
http://dx.doi.org/10.2514/1.B36949.

133

You might also like