Large scale liquid propellant rocket engines such as those used for launch vehicles operate at harsh thermal conditions such as pressures higher than 50 times the atmospheric value and temperatures reaching over 3000K. At these conditions, material durability becomes the limiting factor in engine performance. Design process of liquid rocket engines traditionally relies on intuition and experiments which often yields high costs and less than optimum performance.
Advancement of computational fluid dynamics (CFD) methods for turbulent combustion problems is, on the other hand, promising to aid and eventually dominate the design and analysis of such combustors. However, significant issues related to the CFD analysis of such flows exist to date. There is a community wide effort toward developing and refining the CFD tools for this type of problems.
Our research encompasses
- Systematic evaluation of some of the available CFD methods for turbulent combustion problems,
- Development of new models based on first principles in order to wean CFD simulations off the empirical relations,
- Developing surrogate based approaches for design optimization.
CFD Model Reliability
Many different modeling approaches for turbulent combustion problems are available, ranging in fidelity and flavor. Some popular choices are the finite rate reaction model or the flamelet approach for the chemistry modeling, RANS or LES models for turbulence closure. Of course there are many variations of these models within themselves. The reason for the great variety is that there is no consensus on which model offers the best accuracy or efficiency. Chemistry and turbulence are two challenging problems by themselves that when combined, create an even more elusive problem of their interaction.
In our research, out target is to quantify effects of the spatial and temporal resolutions, chosen chemistry mechanisms, turbulence chemistry interactions and experimental uncertainty.
We take, as a basis, different experimental setups of single element H2/O2 injectors as well as a set of simpler, unit-physics problems representing the break-down of the complex physics involved in full scale simulations. Examples to the unit physics problems include turbulent, non-reacting and reacting coaxial jet mixing problems.
Graduate students: Emre Sozer, Ez Hassan, Seokjun Yun
Representative publications:
- Sozer, E., Vaidyanathan, A., Segal, C., and Shyy, W., “Computational Assessment of Gaseous Reacting Flows in Single Element Injector”, AIAA-2009-449, 47th AIAA Aerospace Sciences Meeting including The New Horizons Forum and Aerospace Exposition, Orlando, Florida, Jan. 5-8, 2009.
- Mack, Y., Haftka, R., Segal, C., Queipo, N., Shyy, W., “Computational modeling and sensitivity evaluation of liquid rocket injector flow,” AIAA-2007-5592, 43rd AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Cincinnati, OH, July, 8-11, 2007.
Multi-Element Liquid Rocket Injector Simulation and Optimization
CFD model reliability studies performed for problems of single-element injectors and the unit-physics type problems provide valuable insight toward making CFD a reliable design tool for real world liquid rocket injectors design problems. Compared to traditional ad-hoc and experiment based design methods, CFD offers substantial quantitative data and the ability to evaluate a much greater range of design configurations.
One of the major challenges facing liquid rockets is the harsh thermal environment in the combustion chamber. A major goal of the liquid rocket injector design is to minimize the combustion length,i.e., faster mixing and burning of fuel and oxidizer. However, the extent to which the combustion length can be reduced is limited by increased local heat flux to the chamber wall resulting in possible material burn-out and crack. Various injector design approaches provide compromises between these two competing objectives.
Surrogate Based Optimization methodology is being utilized to demonstrate the optimization of multiple-injector configurations with design parameters such as:
- Injector-to-injector radial spacing
- Injector-to-injector circumferential spacing
Objective of the optimization is to obtain the best configuration (or a set of best configurations defined by a Pareto front) for maximum performance while maintaining favorable chamber wall thermal conditions.
Graduate students: Emre Sozer, Ez Hassan, Seokjun Yun
Fluid Flow and Heat Transfer through Porous Media for Liquid Rocket Applications
 |
 |
| Rigimesh material | Rigimesh Schematic |
Porous materials are often used for the injector face plate of liquid propellant rocket engines (LPRE). Fuel bleeds through the porous plate to aid in cooling of the injector face by transpiration while helping injection of fuel at the same time. For example, in P&W’s RL10 engine and Space Shuttle Main Engine (SSME), Rigimesh is used. Rigimesh is formed by pressing layers of sintered stainless steel wire meshes. Rigimesh can qualitatively be classified as a dense, non-uniform, fibrous porous media.
In such porous fluid flow problems, direct numerical simulation of the fluid flow accounting for all geometric details is very costly if not impossible. Thus modeling efforts in this area dating back to Darcy’s experimental study in 1856, have mostly aimed at empirically correlating the pore level flow effects to the bulk fluid motion. The most commonly used approach is to add source terms to the global transport equation of mass, momentum and energy and relate these source terms back to empirically found parameters corresponding to the porous material being used.
To achieve predictive capabilities for such porous media flow problems, a first principle-based, multi-scale modeling strategy is developed. The effect of porous media on the macroscopic fluid flow structures is accounted for via local volume averaged governing equations. The resulting set of transport equations, at the global domain level, contains closure terms representing the statistical flow characteristics around the pores. Most porous media can be thought of as a matrix of repeating pore patterns. The closure terms can be directly computed for different flow speeds and pore patterns observed. These results can be interpolated to obtain the closure term evaluations throughout the porous medium. Thus, we can avoid the computational cost of direct simulation yet we can produce accurate numerical predictions completely free of empiricism.
Graduate students: Emre Sozer
Representative publications:
- Sozer, E. and Shyy, W., “Multi-scale thermo-fluid transport in porous media”, International Journal of Numerical Methods for Heat & Fluid Flow, Vol. 18, Issue 7/8, (2008), pp. 883-899, Highly Commended Award Winner at the Literati Network Awards for Excellence 2009.