Definition and Evaluation of Advanced Rocket Thrust Chamber Demonstrator Concepts

The expander cycle demonstrator TCD1 is similar in design to the Ariane 6 new upper stage engine VINCI  [23] and ArianeGroup’s technology precursor for future expander cycle engines [8]. Similarly, the demonstrator TCD1 shows an elongated cylindrical section of the combustion chamber as illustrated in Fig. 1b. As stated above, one objective for the improvement of expander cycle engines is the shortening of this elongated cylindrical part of the combustion chamber in order to decrease the engine mass. However, a sufficient coolant heat up has to be ensured in order to close the engine cycle. Thus, the decreased surface area has to be compensated by an enhanced heat transfer on the hot gas and/or the coolant side requiring an innovative combustion chamber design. One possible solution is an increased roughness on the hot gas surface. In the frame of project K4 a numerical study on wall roughness effects was performed and its results are summarized in Sect. 3.2.

Numerical analyses performed with ArianeGroup’s in-house tool Rocflam3 have been supplemented by a steady-state 3D RANS of project C5 on a 90° segment using laminar finite-rate chemistry. Experimental investigations of increased wall roughness in cooling channels have been carried out in project D9. These experiments have been supplemented with wall-modeled large eddy simulations from project D4 that have in turn been compared to RANS simulations with widespread eddy viscosity models - such as the well known k-\(\varepsilon \) and k-\(\omega \)-SST two equation models - as well as Reynolds stress models that are able to resolve turbulence anisotropies.

The main stage gas generator cycle demonstrator TCD2 aimed at an optimized combustion chamber with reduced system pressure drop trading as this decreases the required pressure head of the turbopump and hence allows a lighter design. However, a significant decrease of the total pressure drop is only possible when the injection pressure drop is reduced below conventionally applied margins and the cooling channel features an increased hydraulic diameter. This in turn makes detailed combustion stability investigations necessary as a coupling of combustion oscillations with the feed system is facilitated and additional cooling methods of the hot gas wall become necessary due to the reduced coolant velocities.

The initially provided design intentionally exhibits hot gas wall temperatures exceeding the maximum tolerable material temperatures. Several projects of the SFB-TRR 40 have proposed technical solutions such as a Ni-based thermal barrier coating designed by project D2 or an innovative cooling concept designed by project A5 utilizing a transpiratively cooled combustion chamber segment in the thermally loaded area manufactured from porous ceramics. As the pressure drop of the injection elements was reduced below common design practice complementary stability investigations have been carried out by projects A4, C3 and C7 and possible countermeasures such as Helmholtz and \(\lambda /4\) resonators have been prepared using high fidelity tools as well as by transfer of experimental results of sub-scale experiments to the demonstrator. The proposed nozzle structure has been evaluated with finite element methods using novel shell elements developed by project D10. A buckling of the structure under thermal and pressure loads resulting both from the hot gas and the outer flow has been investigated. Additionally, the feasibility of an innovative film cooled nozzle extension has been studied by projects A2 and A4 as an alternative to the proposed dump cooled nozzle version. For the development of high order numerical tools such as A1’s porous media solver relevant boundary conditions and operating points have been supplied. State-of-the-art evaporation models proposed by project C4  [4, 10] have been implemented in Rocflam3 and tested extensively.

As alternative fuels have become a major research topic in recent years the main stage gas generator cycle demonstrator TCD3 is designed to be operated with either H\(_\text {2}\)/O\(_\text {2}\) or CH\(_\text {4}\)/O\(_\text {2}\) utilizing an unchanged thrust chamber configuration. Additionally, TCD3 is intended for reuse and thus an increased life is required.

To achieve the necessary high life several projects suggested additional cooling methods similar to those for demonstrator TCD2 such as a dedicated thermal barrier coating proposed by project D2. Thermo-mechanical analyses for the complete life-cycle of TCD3’s combustion chamber structure have been carried out by project D3. For the operation of TCD3 with methane as fuel the accurate prediction of the combustion efficiency and thermal load of the structure is highly important. However, the combustion chemistry of methane is much more complex than that of hydrogen. Hence, well instrumented basic research and sub-scale experiments such as the seven element combustion chamber of project K1 are valuable for the verification of numerical models. Within the SFB-TRR 40 projects K1 and C1/C6 have been working on different combustion models based on non-adiabatic flamelets. A benchmarking of different modeling strategies of groups from the SFB-TRR 40 as well as additional groups from DLR, Jaxa and the University of Harbin has been performed in the frame of the summer program 2017 and published  [16]. A detailed combustion analysis of demonstrator TCD3 has been performed within project C1/C6  [25]. As the investigation of innovative nozzle concepts and advanced methods for aft-body flow control were major research topics of the SFB-TRR 40, comprehensive research on the transition behaviour of dual bell nozzles under real operating conditions, i.e. under consideration of the launcher base flow, has been carried out by division B and project K2.

For early design validations of all three TCDs 2D axisymmetric CHT analyses have been performed. For this purpose Rocflam3 (in 2D/axisymmetric mode) has been coupled to RCFS-II, ArianeGroup’s well-validated 1D in-house engineering tool for the coupled simulation of hot gas and coolant side heat transfer  [14]. The initial design of all three TCDs was validated applying these tools  [5, 6]. In Fig. 2 the hot gas flow fields of these analyses of all three TCDs in \(\text {H}_\text {2}\)/\(\text {O}_\text {2}\) operation are displayed. The right half shows the temperature fields and additional black lines indicate the stoichiometric mixture fraction, i.e. the main reaction zone. The left half illustrates the relative deviation of the local mixture fraction from the globally injected mixture fraction on a logarithmic scale where the area with a relative deviation below 1% is colored in green. All three demonstrators show excellent mixing and a high combustion efficiency \(\eta _{c^*}\) exceeding 99% can be concluded.

In addition to the numerical proof-of-concept for all three demonstrators, axisymmetric simulations have been carried out for further design variations of the expander cycle thrust chamber demonstrator TCD1. The initial design of demonstrator TCD1 exhibits the typical elongated cylindrical section of the combustion chamber like e.g. the European Vinci. This serves to ensure sufficient heat pick-up in the coolant in order to close the expander cycle but leads to a high engine mass. However, performance-wise the necessary characteristic length \(l^*\) is exceeded by far and the combustion chamber could be shortened without a penalty to the combustion efficiency \(\eta _{\text {c}^*}\). Hence, heat transfer enhancement measures are of high interest as they allow a shortening of the cylindrical section of the combustion chamber and thus reducing its mass while ensuring sufficient heat pickup in the coolant.

Axisymmetric CHT analyses of demonstrator TCD1  [6] have shown that a shortened configuration can achieve a sufficient heat pick up by increasing the hot gas wall roughness. However, the increased roughness leads to a considerable rise in the hot gas wall temperature in excess of the maximum allowed wall temperature of the utilized materials (cf. Fig. 3). Hence, a modification of only the hot gas heat transfer is not sufficient and an optimized design has to be elaborated incorporating also structure and coolant side.

For fully 3D CHT investigations the Rocflam3 code was coupled to the commercial solver Ansys CFX. The different simulation domains \(\Omega \) and their interfaces \(\Gamma \) are illustrated in Fig. 4.

Besides the combustion domain \(\Omega _\text {C}\), two solid domains (\(\Omega _\text {S,Cu}\) and \(\Omega _\text {S,Ni}\)) and the domain of the coolant \(\Omega _\text {F}\) had to be simulated. Each domain required its own modeling approach and a treatment of the interfaces to neighboring domains. For the CHT investigations the combustion domain \(\Omega _\text {C}\) was solved with Rocflam3 while Ansys CFX was used for the heat conduction problem in both structure domains (\(\Omega _\text {S,Cu}\) & \(\Omega _\text {S,Ni}\)) and the fluid flow of the coolant (\(\Omega _\text {F}\)). At the interfaces between the domains the heat flux has to be conserved. This is internally ensured by Ansys CFX for the interface \(\Gamma _\text {IF,S-F}\) while the interface \(\Gamma _\text {IF,S-C}\) is handled by an external Python script.

The above summarized method has been applied  [7] to the virtual demonstrator TCD1. The resulting hot gas temperature field as well as selected axial slices of the structure and coolant temperatures of the CHT simulation are illustrated in Fig. 5. Individual reaction zones in the wake of each injection element can be identified by the shown stoichiometric surfaces. They can be interpreted as “flames” and exhibit a uniform behavior except for the outermost row. Here an elongated flame shape can be recognized. Numerical investigations  [17] in a sub-scale combustion chamber show a similar behavior of Menter’s SST model whereas simulations with a \(k-\varepsilon \) turbulence model led to flames of uniform length. The authors explain the increased flame length of the SST model with a reduced mixing due to decreased turbulence production in the area of active \(k-\omega \) model. Nevertheless, in axial direction a fast homogenization of the temperature can be observed indicating an overall good mixing process and hinting at a high combustion efficiency.

The resulting hot gas wall temperature profile is depicted in Fig. 6. The lower plot shows the axial wall temperature profile whereas the upper contour plot shows the circumferential temperature variation over the wall angle \(\Phi =\arctan \left( z/y\right) \). Black dashed lines indicate the position of the injection elements of the outermost row at \(-15\)°, \(-3\)° and 9°. In the throat (\(x=0\text {m}\)) a local circumferential temperature minimum in the wake of the injection elements can be observed due to locally increased hydrogen mass fractions. Furthermore, a locally reduced wall temperature can be noticed at the positions of the cooling channels, most pronounced at around \(x=-0.1 \text {m}\). Additional stratification due to local condensation of water can be seen in the supersonic part of the combustion chamber. The circumferential temperature stratification reaches a magnitude of up to 100 K which has to be considered in liner life analyses.

However, in addition to the circumferential stratification liquid rocket engine combustion chambers are also subjected to a high thermal gradient in radial direction due to the strong regenerative cooling. This can be recognized when Fig. 7 is considered where the local temperatures at several radial planes are plotted. The position of each plane is indicated in the sketch on the right where an axial section of a single cooling channel and fin is displayed. Starting with the strongly stratified hot gas wall temperature profile (a) already a significant temperature drop towards the cooling channel bottom wall (b) can be noticed. The circumferential stratification however remains nearly constant. Towards the upper cooling channel wall (c) and the copper-nickel interface (d) the structure temperature is reduced drastically and nearly reaches the coolant bulk temperature level. The circumferential stratification is strongly reduced.

Considering the coolant bulk temperature only a minor variation of about 1% can be noticed between the individual cooling channels. As in full-scale tests most experimental data on structure temperatures is obtained either on the combustion chamber’s outer wall or derived from few local thermocouples integrated in the wall a precise prediction of the maximum occurring hardware temperatures from experimental data is not possible. Well anchored numerical simulations can thus supplement experimental data and provide further insight.