With advances in computational speed, computing power, and the affordability of digital processors in the last two decades, aeronautics researchers in industry, academia, and the government have turned increasingly to computational simulations to model complex physical and chemical phenomena. Industry is motivated by the possibility of using computational simulations to reduce the cost and time of product development, while increasing product reliability. Academic and government researchers also value the ability to attack more complex problems. These computational simulations generally employ a number of physics-based models within the governing conservation and state equations. Examples include models that describe droplet behavior and interactions, particulate matter formation, turbulence, turbulence–chemistry interactions, boundary-layer growth and transition, fracture, crack propagation, and material phase boundaries. A lack of fundamental understanding often requires these models to contain adjustable parameters that are grossly calibrated to empirical data sets. These data sets are often incomplete, which means that untested assumptions must be incorporated in the models. The computational simulations are generally not validated in spatial detail except for comparison of code predictions to input and output measurements. Additionally, the adjustable parameters are tweaked to match predictions with measurements. It is not uncommon to find that the codes do not extrapolate well when the design space changes considerably, prompting more tweaking of the adjustable parameters. Also, when results are presented, details are usually omitted in connection with the use of boundary conditions or how adjustable parameters were set, making it harder for independent researchers to reproduce the results. Thus to a large extent, empiricism has transitioned from the physical to the computational realm, but it persists. Nevertheless, within their applicable ranges, computational simulations have enabled technical progress, as witnessed by the state of aeronautics today. Unfortunately, limits on the use of simulations are often not well understood.
NASA and its academic and industrial partners can make very significant contributions in developing and validating physics-based analysis tools. These are readily assimilated by industry into their proprietary product design codes. NASA, industry, and academia can jointly participate in research into physics-based analysis tools because it is fundamental in nature, publishable, and sharable. This research will take time to mature, yet advances can readily be translated into practice as they occur. Furthermore, given the budget- and schedule-driven nature of the aerospace business, this is the type of work that industry can no longer afford to pursue. Developing physics-based tools whose accuracy and range of applicability limits are well established is a lengthy, iterative process. Validation requires well-designed experiments to elucidate the underlying physics as well as experimental facilities of appropriate scale and advanced, nonperturbing diagnostics to perform detailed, spatially and temporally resolved measurement of parameters.
Advances in physics-based analysis tools would help address R&T Challenges in all of the R&T Areas. For example, turbulence modeling is a key element in the accurate prediction of mixing, which is very important in many aspects of aerodynamics (A2), aeroacoustics (A4a, B1a), and combustion processes (B1b). Accurate predictions of flow separation are a prerequisite to the successful design of both nonreacting (A2, A4a, A4b) and reacting (B1a, B1b, B5, B8) flow devices. Mathematical models of material properties and reactions are essential for structural response (A4a, C4, C10). Droplet–droplet and droplet–flow interactions are important processes in predicting both icing (A6) and combustion behavior (B1b, B8, B10). Modeling flow–structure interactions accurately is an important element of aeroelasticity and noise generation (A4a, B1a, C5, C6b). The development of higher-temperature alloys is key to improving propulsion system fuel efficiency (B4, C6a). Many of the computational science issues associated with large, complex computational simulations, such as automated grid generation, parallelizing codes, and error propagation analyses, are common elements across several R&T Areas.
The following R&T Challenges would benefit significantly from using physics-based analysis tools: A1, B1a, B1b, E1, A2, E2, D3, A4a, A4b, B4, C4, D4, E4, B5, C5, A6, B6a, C6a, C6b, A7b, B8, A9, B10, and C10.
Discipline-specific design tools, including optimization and inverse design, have improved the performance of airfoils, wings, structures, control systems, and propulsion systems for many years, and they are now critical parts of the design process. The next step in the design of more complex systems involves more than just combining these disciplinary tools or gluing together discipline-specific analyses and optimization. New multidisciplinary tools are needed to integrate high-fidelity analyses with efficient design methods and to accommodate uncertainty, multiple objectives, and large-scale systems. Research in efficient methods for including large numbers of design variables (e.g., adjoint methods, multifidelity models, and surrogate models), probabilistic design methods, and tools for distributed, complex systems is particularly important to the development of future aeronautical systems.