Abstract
This paper studies design problems where the performance is dominated by the dynamic evolution of interfaces due to chemical processes. Considering the representative example of a solid rocket motor, the shape of the interface between the solid fuel and the gas inside the combustion chamber at the beginning of the burn process and the reference burn rate of a functionally graded propellant are optimized to achieve a desired thrust over time. The initial fuel–gas interface is described by a level set function parameterized by geometric primitives and B-splines. The reference burn rate distribution is discretized by multi-variate B-splines. The thrust is predicted by a semi-analytical approach that requires modeling the recession of the fuel–gas interface. To this end, a stabilized finite element formulation of the Hamilton–Jacobi equation is used to describe the evolution of the level set function during the burn process. The optimization problem is solved by a nonlinear programming method, and the design sensitivities are evaluated by the adjoint method. The proposed optimization approach is studied with numerical examples in 2D and 3D, involving configurations with more than \(6 \times 10^{4}\) optimization variables and \(12 \times 10^{6}\) state variables. The optimization results show that this approach provides a promising design tool for problems with dynamically evolving interfaces due to surface reactions. However, the results also reveal that the simplicity of the recession and thrust models requires limiting the design freedom through a carefully chosen design parameterization. Furthermore, additional constraints need to be imposed to prevent unphysical designs.
Similar content being viewed by others
References
Albarado KM, Hartfield RJ, Wade Hurston B, Jenkins RM (2011) Solid rocket motor performance matching using pattern search/particle swarm optimization. In: 47th AIAA/ASME/SAE/ASEE joint propulsion conference and exhibit 2011. https://doi.org/10.2514/6.2011-5798
Albarado K, Shelton A, Hartfield R (2012) SRM simulation using the level set method and higher order integration schemes. In: 48th AIAA/ASME/SAE/ASEE joint propulsion conference and exhibit, p 4215
Barrère M, Jaumotte A, Fraeijs de Veubeke B, Vandenkerckhove J (1960) Rocket propulsion. Elsevier, Amsterdam
Barth TJ, Sethian JA (1998) Numerical schemes for the Hamilton–Jacobi and level set equations on triangulated domains. J Comput Phys 145(1):1–40
Bazilevs Y, Calo VM, Tezduyar TE, Hughes TJR (2007) YZ\(\beta\) discontinuity capturing for advection-dominated processes with application to arterial drug delivery. Int J Numer Methods Fluids 54(6–8):593–608. ISSN 02712091. https://doi.org/10.1002/fld.1484
Bernardini M, Cimini M, Stella F, Cavallini E, Di Mascio A, Neri A, Martelli E (2020) Large-eddy simulation of vortex shedding and pressure oscillations in solid rocket motors. AIAA J 58(12):5191–5201. ISSN 1533385X. https://doi.org/10.2514/1.J058866
Bertin JJ, Cummings RM (2006) Critical hypersonic aerothermodynamic phenomena. Annu Rev Fluid Mech 38:129–157. ISSN 00664189. https://doi.org/10.1146/annurev.fluid.38.050304.092041
Blazek J (2003) Flow simulation in solid rocket motors using advanced CFD. In: 39th AIAA/ASME/SAE/ASEE joint propulsion conference and exhibit, 2003. https://doi.org/10.2514/6.2003-5111
Candler GV (2019) Rate effects in hypersonic flows. Annu Rev Fluid Mech 51:379–402. ISSN 00664189. https://doi.org/10.1146/annurev-fluid-010518-040258
Chandru RA, Balasubramanian N, Oommen C, Raghunandan BN (2018) Additive manufacturing of solid rocket propellant grains. J Propuls Power 34(4):1090–1093. ISSN 15333876. https://doi.org/10.2514/1.B36734
Davenas A (2012) Solid rocket propulsion technology. Pergamon Press, Inc., Tarrytown
Fabignon Y, Anthoine J, Davidenko D, Devillers R, Dupays J, Gueyffier D, Hijlkema J, Lupoglazoff N, Lamet JM, Tessé L, Guy A, Erades C (2016) Recent advances in research on solid rocket propulsion. AerospaceLab 11:15
Falgout RD (2006) An introduction to algebraic multigrid. Comput Sci Eng 8(6):24–33. ISSN 15219615. https://doi.org/10.1109/MCSE.2006.105
Fuller JK, Ehrlich DA, Lu PC, Jansen RP, Hoffman JD (2011) Advantages of rapid prototyping for hybrid rocket motor fuel grain fabrication. In: 47th AIAA, ASME, SAE, ASEE joint propulsion conference and exhibit, 2011. ISBN 9781600869495. https://doi.org/10.2514/6.2011-5821
Gale MP, Alam MF, Mehta RS, Luke EA (2020) Multiphase modeling of solid rocket motor internal environment. In: AIAA Propulsion and Energy 2020 Forum, pp 1–12. ISBN 9781624106026. https://doi.org/10.2514/6.2020-3934
Gueyffier D, Roux FX, Fabignon Y, Chaineray G, Lupoglazoff N, Vuillot F, Alauzet F (2014) High-order computation of burning propellant surface and simulation of fluid flow in solid rocket chamber. In: 50th AIAA, ASME, SAE, ASEE joint propulsion conference, 2014. ISBN 9781624103032. https://doi.org/10.2514/6.2014-3612
Hartfield R, Jenkins R, Burkhalter J, Foster W (2003) A review of analytical methods for solid rocket motor grain analysis. In: 39th AIAA/ASME/SAE/ASEE joint propulsion conference and exhibit, 2003. ISBN 9781624100987. https://doi.org/10.2514/6.2003-4506
Hartfield R, Jenkins R, Burkhalter J, Foster W (2004) Analytical methods for predicting grain regression in tactical solid-rocket motors. J Spacecr Rockets 41(4):689–693
Hejl RJ, Heister SD (1995) Solid rocket motor grain burnback analysis using adaptive grids. J Propuls Power 11(5):1006–1011. ISSN 07484658. https://doi.org/10.2514/3.23930
Kamran A, Guozhu L (2012) An integrated approach for optimization of solid rocket motor. Aerosp Sci Technol 17(1):50–64
Lambert JD (1993) Numerical methods for ordinary differential systems: the initial value problem. Math Comput 61(204):939. ISSN 00255718. https://doi.org/10.2307/2153266
Li Q, Qiang He G, Jin Liu P, Li J (2014) Coupled simulation of fluid flow and propellant burning surface regression in a solid rocket motor. Comput Fluids 93:146–152. ISSN 00457930. https://doi.org/10.1016/j.compfluid.2014.01.028
Losasso F, Fedkiw R, Osher S (2006) Spatially adaptive techniques for level set methods and incompressible flow. Comput Fluids 35(10):995–1010
Mahjub A, Mazlan NM, Abdullah MZ, Azam Q (2020) Design optimization of solid rocket propulsion: a survey of recent advancements. J Spacecr Rockets 57(1):3–11. ISSN 15336794. https://doi.org/10.2514/1.A34594
McClain MS, Gunduz IE, Son SF (2019) Additive manufacturing of ammonium perchlorate composite propellant with high solids loadings. Proc Combust Inst 37(3):3135–3142. ISSN 15407489. https://doi.org/10.1016/j.proci.2018.05.052
Moon P, Spencer DE (1961) The vector Helmholtz equation. In: Field theory handbook, pp 136–143. https://doi.org/10.1007/978-3-642-83243-7_5
Muravyev NV, Monogarov KA, Schaller U, Fomenkov IV, Pivkina AN (2019) Progress in additive manufacturing of energetic materials: creating the reactive microstructures with high potential of applications. Propellants Explos Pyrotech 44(8):941–969. ISSN 15214087. https://doi.org/10.1002/prep.201900060
Najjar FM, Haselbacher A, Ferry JP, Wasistho B, Balachandar S, Moser RD (2003) Large-scale multiphase large-eddy simulation of flows in Solid-Rocket motors. In: 16th AIAA computational fluid dynamics conference, 2003. ISBN 9781624100864. https://doi.org/10.2514/6.2003-3700
Olsson E, Kreiss G (2005) A conservative level set method for two phase flow. J Comput Phys 210(1):225–246
Osher S, Fedkiw RP (2003) Level set methods and dynamic implicit surfaces. Springer, New York
Osher S, Sethian JA (1988) Fronts propagating with curvature-dependent speed: algorithms based on Hamilton–Jacobi formulations. J Comput Phys 79(1):12–49
Peterson E, Nielsen G, Johnson W (1968) Generalized coordinate grain design and internal ballistics evaluation program. In: 3rd Solid propulsion conference, 1968. https://doi.org/10.2514/6.1968-490
Raza MA, Liang W (2013) Design and optimization of 3D wagon wheel grain for dual thrust solid rocket motors. Propellants Explos Pyrotech 38(1):67–74
Rumpfkeil MP, Zingg DW (2010) The optimal control of unsteady flows with a discrete adjoint method. Optim Eng 11(1):5–22. ISSN 13894420. https://doi.org/10.1007/s11081-008-9035-5
Saad Y (2003) Iterative methods for sparse linear systems. SIAM. https://doi.org/10.1137/1.9780898718003
Sachdev JS, Groth CP, Gottlieb JJ (2005) Parallel AMR scheme for turbulent multi-phase rocket motor core flows. In: 17th AIAA computational fluid dynamics conference, 2005. ISBN 9781624100536. https://doi.org/10.2514/6.2005-5334
Sethian JA (1990) Numerical algorithms for propagating interfaces: Hamilton–Jacobi equations and conservation laws. J Differ Geom 31(1):131–161
Sethian JA (1996) Theory, algorithms, and applications of level set methods for propagating interfaces. Acta numer 5:309–395
Sethian JA (1999) Level set methods and fast marching methods: evolving interfaces in computational geometry, fluid mechanics, computer vision, and materials science, 2nd edn. Cambridge University Press, Cambridge
Sethian JA, Smereka P (2003) Level set methods for fluid interfaces. Annu Rev Fluid Mech 35(1):341–372
Sullwald W, Smit GJ, Steenkamp AJ, Rousseau CW (2013) Solid rocket motor grain burn back analysis using level set methods and Monte Carlo volume integration. In: 49th AIAA/ASME/SAE/ASEE joint propulsion conference, 2013. AIAA, p 4087
Sutton GP, Biblarz O (2001) Rocket propulsion elements. Wiley. ISBN 0-471-32642-9. https://doi.org/10.1016/0016-0032(49)90439-1
Svanberg K (2002) A class of globally convergent optimization methods based on conservative convex separable approximations. SIAM J Optim 12(2):555–573. ISSN 10526234. https://doi.org/10.1137/S1052623499362822
Tezduyar TE, Park YJ (1986) Discontinuity-capturing finite element formulations for nonlinear convection–diffusion–reaction equations. Comput Methods Appl Mech Eng 59(3):307–325. ISSN 00457825. https://doi.org/10.1016/0045-7825(86)90003-4
Tola C, Nikbay M (2018) Solid rocket motor propellant optimization with coupled internal ballistic-structural interaction approach. J Spacecr Rockets 55(4):936–947. ISSN 15336794. https://doi.org/10.2514/1.A34066
Touré MK, Soulaïmani A (2016) Stabilized finite element methods for solving the level set equation without reinitialization. Comput Math Appl 71(8):1602–1623. ISSN 08981221. https://doi.org/10.1016/j.camwa.2016.02.028
Vandenbrande J (2020) DARPA TRADES challenge problems overview. http://solidmodeling.org/trades-cp/. Accessed 27 Nov 2021
Wang X, Jackson T, Massa L (2004) Numerical simulation of heterogeneous propellant combustion by a level set method. Combust Theory Model 8(2):227
Wang D-H, Fei Y, Hu F, Zhang W-H (2014) An integrated framework for solid rocket motor grain design optimization. Proc Inst Mech Eng G 228(7):1156–1170
Willcox MA, Brewster MQ, Tang KC, Stewart DS (2007) Solid propellant grain design and burnback simulation using a minimum distance function. J Propuls Power 23(2):465–475. ISSN 15333876. https://doi.org/10.2514/1.22937
Wu Z, Wang Z, Zhang W, Okolo PN, Fei Y (2017) Solid-rocket-motor performance-matching design framework. J Spacecr Rockets 54(3):698–707. ISSN 15336794. https://doi.org/10.2514/1.A33655
Wu Z, Wang D, Wang W, Okolo PN, Zhang W (2019) Solid rocket motor design employing an efficient performance matching approach. Proc Inst Mech Eng G 233(11):4052–4065. ISSN 20413025. https://doi.org/10.1177/0954410018814037
Yildirim C, Aksel H (2005) Numerical simulation of the grain burnback in solid propellant rocket motor. In: 41st AIAA/ASME/SAE/ASEE joint propulsion conference and exhibit, 2005. https://doi.org/10.2514/6.2005-4160
Yücel O, Açık S, Toker KA, Dursunkaya Z, Aksel MH (2015) Three-dimensional grain design optimization of solid rocket motors. In: 2015 7th International conference on recent advances in space technologies (RAST), 2015. IEEE, pp 471–476
Acknowledgements
The authors acknowledge the support of the Defense Advanced Research Projects Agency (DARPA) under the TRADES program (Agreement HR0011-17-2-0022). The opinions and conclusions presented in this paper are those of the authors and do not necessarily reflect the views of DARPA.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Replication of results
The TO and XFEM calculations were performed using an in-house solver that is not at the stage of being publicly available.
Additional information
Responsible Editor: James K. Guest
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Topical Collection: 14th World Congress of Structural and Multidisciplinary Optimization.
Guest Editors: Y. Noh, E. Acar, J. Carstensen, J. Guest, J. Norato, P. Ramu.
Appendices
Appendix 1: Verification example for Hamilton–Jacobi transport model
To verify the accuracy of the HJ transport model presented in Sect. 3.1 and to identify important model and simulation parameters, we consider the evolution of a level set field with an initially circular zero isocontour in 2D and an initially spherical zero isocontour in 3D for a uniform speed field. For these problems, trivial analytical solutions exist with the radii increasing linearly in time.
The configurations are shown in Fig. 28. In the 2D case, the computational domain is a square of \(0.2 \times 0.2\) m. The initial solid–gas interface is circular. In the 3D case, the computational domain is a cube of \(0.1 \times 0.1 \times 0.1\) m. The level set function is initialized to describe a spherical solid–gas interface. In both cases, the level set field evolves in time due to a uniform speed field with a reference burn rate of 0.005 m/s. Thus, the solution of the speed field extension model is constant across the computational domain.
With these configurations, we study the accuracy of the simulation with respect to mesh size, time step size, and the algorithmic parameters defined in Sect. 3.1. Note that the dimensions and burn rates are representative of the rocket engine configurations studied in Sects. 6 and 7. The accuracy of the numerical simulations is measured by comparing the surface area of the solid–gas interface against that of the analytical solution. In the finite element model, the interface area is computed as follows:
where the smoothed Dirac function, \(f_{\text {drc}}\), is computed by (17).
For the 2D case, the baseline configuration uses a \(128 \times 128\) mesh and a time step size of \(\Updelta t=0.05\) s. For the 3D case, the baseline configuration uses a \(25 \times 25 \times 25\) mesh and a time step size of \(\Updelta t = 0.1\) s. In both cases, the sharpness parameter to compute the interface area with (46) and (17) is \(\lambda _{\text {drc}}=1/h_{\text {ele}}\). The Eikonal stabilization is set to \(\beta _{\text {eik}}=1.0\) and a uniform target gradient norm of \(\Vert \varphi \Vert _{\text {trg}}=1.0\) is used.
The influences of the mesh size, time step size, the sharpness parameter \(\lambda _{\text {drc}}\), and the Eikonal stabilization \(\beta _{\text {eik}}\) on the accuracy of the solid–gas interface area prediction are studied. The numerical results suggest that for the circular and spherical geometries considered here the influence of \(\beta _{\text {eik}}\) is negligible. In contrast, the other parameters have a noticeable influence on the numerical results, which are summarized in Table 2 for the 2D case and Table 3 for the 3D case.
These results show that a sufficient mesh refinement is crucial to obtain accurate results. This effect can be clearly seen from the 3D mesh refinement study, where the baseline configuration has a rather coarse mesh. For the examples considered here, the dependence on the time step size is small and the error is dominated by the mesh resolution error. For problems with solid–gas interfaces merging during the burn process, a stronger dependence of the time step size on the accuracy was observed. This study also illustrates the importance of using a proper sharpness parameter for the smoothed Dirac function. The error can significantly increase if the sharpness parameter is too large. This effect is even more pronounced if more complex shapes are considered.
Appendix 2: Verification example for detecting disconnected volumes of gas
The ability of the approach presented in Sect. 3.2 to detect volumes of gas disconnected from the nozzle is illustrated with an example shown in Fig. 29. The computational domain of size \(100 \times 100 \times 633\) mm is discretized by a \(15 \times 15 \times 95\) uniform mesh. This results in an element size of \(h_{\text {ele}} = 6.67\) mm. The reference value is set to \({{\bar{T}}}_{1} = 100\), the source value to \(T_{1}^{\text {src}} = -100\), the convection coefficient parameter to \(\alpha _{T1} = 10^{ -4}\), and the sharpness parameter to \(\lambda _{\text {dif}} = 5/h_{\text {ele}}\) . The cylindrical volumes of gas have a diameter of 10 mm. The gap between the cylinders is placed at one and two thirds of the \(x_{3}\)-dimension of the computational domain.
First, a gap size of 25 mm is considered. The scalar field contour plot on the right side of Fig. 29 shows that for this gap size the scalar field value in the volume of gas connected to the nozzle is approximately \(T_{1} \approx 100\), while the values in the volumes of gas disconnected from the nozzle are \(T_{1} \approx - 100\). Thus, using the scalar field values, one can easily distinguish between volumes of gas connected to the nozzle and disconnected from the nozzle.
In Fig. 30, we show scalar field contour plots for varying gap sizes on a slice through the center of the computational domain in the \(x_{1}-x_{3}\) plane. The gap size is altered by elongating the axial length of the cylindrical gas volumes. The results indicate that one can distinguish between connected and disconnected volumes of gas until the gap size is approximately smaller than two times the element size.
Appendix 3: Verification example for detecting unburnt propellant
Figure 31 shows a numerical example with two isolated spherical pieces of unburnt fuel within a volume of fuel that is connected to the casing. As in the example of Section B, the computational domain of size \(100 \times 100 \times 633\) mm is discretized by a \(15 \times 15 \times 95\) uniform mesh. Here, the reference value is set to \({{\bar{T}}}_{2} = - 100\), the source value to \(T_{2}^{\text {src}} = 100\), the convection coefficient parameter to \(\alpha _{T1} = 10^{ -4}\), and the sharpness parameter to \(\lambda _{\text {dif}} = 5/h_{\text {ele}}\). The spherical pieces of propellant have a radius of 15 mm and are contained within a cylindrical volume of gas with a radius of 60 mm. The solution of the auxiliary diffusion problem allows the identification of the isolated piece of unburnt fuel. Their scalar field value is \(T_{2} \approx 100\), while the scalar field in the fuel connected to the casing is \(T_{2} \approx -100.\)
To study the resolution of this model, we vary the radius, \(R_{\text {inner}}\), of the spherical piece of unburnt fuel. The radius, \(R_{\text {outer}}\) of the outer cylindrical cavity remains constant. In Fig. 32, the scalar field values are shown on a slice through the center of the upper sphere for increasing radii. The circular solid lines represent the solid–gas interfaces. These results suggest that one can identify pieces of unburnt fuel if the distance between fuel connected to and disconnected from the casing is more than twice the element size of the mesh. This observation is consistent with the one for the resolution for detecting connected and disconnected volumes of gas in “Appendix 2”.
Appendix 4: Refined axial parameterization of 3D optimization problem
To explore the influence of refining the parameterization in axial direction, we consider a 4-petal configuration with a radial B-spline discretization of the burn rate field, similar to the 8-petal configuration studied in Sect. 7.3.1. However, we increase the number of B-spline basis functions in axial direction from 23 to 63. To capture the large spatial variability in axial direction, we also refine the mesh in axial direction and use \(40 \times 40 \times 120\) elements, yielding 406,802 level set and speed field degrees of freedom.
The optimization results are shown in Figs. 33 and 34. The thrust profile is given in Fig. 35. Comparing the optimized design with the one obtained with a coarser axial parameterization in Sect. 7.3.1 suggests that allowing for a larger spatial variability in axial direction leads to a distinctly different geometry of the initial solid–gas interface. Instead of forming distinct petals, the initial interface area is increased by an undulation in axial direction. However, the thrust profile of the optimized design does not match the target thrust.
Rights and permissions
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Maute, K., De, S. Shape and material optimization of problems with dynamically evolving interfaces applied to solid rocket motors. Struct Multidisc Optim 65, 218 (2022). https://doi.org/10.1007/s00158-022-03331-9
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00158-022-03331-9