**6. Conclusions**

This work described the multidisciplinary design of an aerodynamic drag device used to allow a passive re-entry, i.e., avoiding retropropulsion, of a reusable launch vehicles' first stage. The drag device consists of four sub-components and represents, in a closed configuration, the interstage of the launcher. In the extended configuration, high thermal and mechanical loads are experienced. To achieve a lightweight design, a holistic design approach is required. Therefore, in this work, both thermal and mechanical analyses are conducted. Three different concepts are compared: One is based on an ablative thermal protection system and a CFRP-aluminium honeycomb sandwich structure. The second is a sandwich structure representing a so-called integrated thermal protection system based on a ceramic matrix composite. The third is, as well, an integrated thermal protection system, whose design is based on the use of metallic lattice structures in which a phase change material is embedded. The main results as well as the outlook for each analysed technology are summarised as follows:

	- **–** The separation of thermal and structural functions allows one to use efficient materials and construction methods for each absolved function, namely PICA for thermal protection and CFRP-aluminium honeycomb sandwich for loadbearing functionality.
	- **–** The solution delivers the lowest overall mass.
	- **–** It is easier to obtain a feasible solution because of the two high-TRL solutions used in this concept.
	- **–** Reusability is a concern. Indeed, after-flight maintenance operations should include either a check of the receded amount of ablative material or a re-application. Alternatively, a fast-swap concept can be considered, directly removing and substituting both the structural element and the thermal protection system.
	- **–** The concept represents a lightweight, reusable solution for thermal protection purposes.
	- **–** However, the thermally optimised solution does not withstand the thermomechanical loads.
	- **–** Although ceramic matrix composites exhibit a low coefficient of thermal expansion, the high thermal gradients and the high stiffness lead to high thermal stresses compared to the low tensile strength of the material. Improvements in this direction are needed to allow a load bearing functionality of CMC-based TPS. Three-dimensional CTE tailoring via appropriate fibre orientation can be considered in future work.
	- **–** The integration of a PCM drastically reduces outer wall (top face sheet) temperatures and therefore allows use of materials with high specific mechanical properties, i.e., Inconel.
	- **–** However, thermal stresses above the yield strength of the respective materials in the different layers are identified. These can be caused by mismatch in the CTE of

the different materials and high bending stiffness. Additionally, the use of copper alloy, although beneficial to improving the thermal conductivity of the PCM, has the drawback of a low specific yield strength.


For use in a reusable microlauncher, a holistic assessment of load-bearing TPS structures is required. Specifically, the reusability requirement could make the use of ablative TPS expensive compared to heavier solutions with lower expected overhaul time and cost between launches.

Future work should aim at improved thermal analyses with better estimation of the boundary conditions, i.e., a better definition of the ambient temperature for the radiative heat exchange term of the outer surface. Such ambient temperature should be based on piecewise interpolation of ambient temperatures at different points during the flight trajectory.

Furthermore, future activities will concentrate on the multi-objective (thermal and mechanical) optimisation of the two reusable TPS solutions (CMC corrugated and lattice PCM). Only in this way can an integrated structure with good thermal and mechanical performance be obtained. More adequate material choice and combination should be considered among the parameters of the optimisation as well. Furthermore, manufacturing constraints that hinder the construction methodology need to be taken into account. Indeed, the manufacturing of CMCs is still not mature enough to monolithically realise such wide and complex components. On the other hand, the maximum size of realisable metallic structures via additive manufacturing is still small compared to the size of the component considered in this work. The joining techniques, e.g., brazing or laser welding, of different parts of the hierarchical sandwich structure may represent a bottleneck and should be thoroughly investigated. Furthermore, compatibility of the chosen PCM with the core and face sheet material combination should be evaluated case by case. The volume expansion of the PCM after melting should also be taken into account. Although technical solutions like the use of membranes or expansion chambers exist, these might affect the overall structural design. Finally, different kinds of unit cells and local tailoring of the cell parameters of lattice structures can be used to obtain a tailored coefficient of thermal expansion. This would allow one to reduce overall thermal stresses. For the high flexibility in the design process, additively manufactured lattice structures can be considered viable candidates to obtain holistically optimised structures with thermal protection functionality.

**Author Contributions:** Conceptualization, S.P.; Methodology, S.P.; Software, N.H. and S.P.; CFD analyses, A.Z.; Thermal analyses, S.P.; Mechanical analyses, D.P. and S.P.; Writing—original draft preparation, S.P. and D.P.; Project administration, K.-U.S., R.B., and A.D.; Funding acquisition, A.D. and R.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by EU Horizon 2020 Programme, grant number 870340.

**Data Availability Statement:** The open source software Hot-STARSHIP is available at the following link: https://github.com/nilsh7/Hot-STARSHIP/tree/v1.0.0 (accessed on 18 of March 2023) or after contacting the corresponding author.

**Acknowledgments:** The authors would like to thank Giovanni Medici from DEIMOS SPACE S.L.U for the information supplied about mission analysis and trajectory. The authors would also like to thank Tobias Schalm and Maximilian Schirp-Schoenen for their redactional help in reviewing the paper.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**

The following abbreviations are used in this manuscript:


#### **Appendix A. Verification of the Hot-STARSHIP Solver**

In the test, a 5 cm thick piece of TACOT material is heated for one minute and cooled off for another minute afterwards. The parameters for this problem are depicted in Table A1.

**Table A1.** Parameters for ablative test.


The aerodynamic boundary condition is used with time-varying values of the transfer coefficient *ρeueCH*<sup>0</sup> and recovery enthalpy *hr* to achieve the heating and cooling phase. Note that, in contrast to the use of pre-generated *B* -tables in [33], the results presented here are computed with our own *B* -tables that are extracted from Mutation++ as part of the process. The time step is chosen to be 0.1 s, and the grid has a first cell thickness of 0.05 mm and a maximum growth factor of 1.03. The temperature and recession history for a calculation with Amaryllis and PATO are obtained from [33]. Figure A1 shows a comparison of surface recession *s* and char and gas mass flow rate (*m*˙ *<sup>c</sup>* and *m*˙ *<sup>g</sup>*) as a function of time.

**Figure A1.** Comparison of Hot-STARSHIP and Amaryllis/PATO recession and mass flow rates.

Hot-STARSHIP analysis results in a 7.6% higher final recession value *s* (15.6 mm versus 14.5 mm). This difference can be attributed to the initial difference in char mass flow rate *m*˙ *<sup>c</sup>* where Hot-STARSHIP peaks, whereas PATO/Amaryllis show a smoother transient behaviour. As time progresses, the two curves approach each other. In the end, the difference in recession rate *s*˙ is about 4.6%. Once the heat flux input ends, both programs conform to each other. The difference in recession amount can also be observed in the temperature plots in Figure A2. The temperatures are plotted in stationary locations. Thus, once the surface has receded to a fixed location, the location's temperature history ends and merges with the surface temperature history at that point. Both programs are in good agreement of the surface temperature. Wider differences are only observable in the first 20 s where the higher char ablation rate of Hot-STARSHIP provides more cooling. In the fixed locations, the temperature difference between both programs grows with time. As noted, part of this is because of the higher recession amount of Hot-STARSHIP. Because with higher recession amounts fixed locations are closer to the surface and the temperature gradients are large due to low conductivity, differences are observed (see also Figure A3).

Note that one of the main constraints of thermal protection system thickness, the backface temperature at 50 mm, is in good agreement for both programs. The final difference is about 12 K, and the average difference is even lower as the two curves cross each other.

**Figure A2.** Comparison of Hot-STARSHIP and Amaryllis/PATO temperature curves.

Finally, Figure A3 shows internal temperature profiles shortly after the heating begins (2 s), when the heating stops (60 s), shortly after the heating stops (60.1 s) at the very end of the calculation (120 s), and some intermediate values.

**Figure A3.** Temperature profiles of Hot-STARSHIP calculation at selected times.

During heating, large temperature gradients are present in the first few millimeters to centimeters, reaching values of up to 700 K/mm. This explains the seemingly large differences in Figure A2 arising from different proximity to the surface. Once the heating ends, the surface temperature drops rapidly, leading to peak temperature not at the surface, but 5 mm into the material. From this point onward, the temperature profile flattens out as dictated by the second order conductivity equation.

The differences between Hot-STARSHIP and PATO or Amaryllis might be attributed to the use of pre-generated *B* -tables for PATO and Amaryllis, whereas this is not the case for Hot-STARSHIP, where the tables are computed via Mutation++. In addition, PATO and Amaryllis are "type 2" [42] solvers, whereas Hot-STARSHIP can be classified as a "type 1" solver. This means that in addition to the details resolved in Hot-STARSHIP, PATO and Amaryllis consider Darcy's law for convective transport of pyrolysis gas as well as porosity and permeability for diffusive transport [42]. Whereas Hot-STARSHIP assumes that the gas leaves instantly, solving Darcy's law as done in PATO could hold back some gas that then flows out more slowly, leading to a higher gas mass flow rate. This behaviour would also increase the cooling of the in-depth material, which explains the lower predicted temperatures of PATO and Amaryllis. On the other hand, the slightly lower surface temperature of Hot-STARSHIP can be explained by the higher char mass flow rate that provides more surface cooling.

### **Appendix B. Material Data**

The following material data were acquired from the Ansys® database for a T700 CFRP composite material and for an aluminium honeycomb.


**Table A2.** UD composite stiffness properties [MPa].

**Table A3.** UD composite failure stresses [MPa].



**Table A4.** Al honeycomb stiffness properties [MPa].

The material data for the CMC are reported with reference to [10]:

**Table A5.** CMC material properties.

