Aerodynamic design of integrated propulsion–airframe configuration of a hybrid wing body aircraft

Abstract

A hybrid wing body (HWB) concept is being considered by NASA as a potential subsonic transport aircraft that meets aerodynamic, fuel, emission, and noise goals in the time frame beyond 2035. While the concept promises advantages over a conventional wing-and-tube aircraft, it poses unknowns and risks, thus requiring in-depth and broad assessments. Specifically, the configuration entails a tight integration of the airframe and propulsion geometries; the aerodynamic impact has to be carefully evaluated. With the propulsion nacelle installed on the (upper) body, the lift and drag are affected by the mutual interference effects between the airframe and nacelle. The static margin for longitudinal stability is also adversely changed. In the present paper, a design approach is developed in which the integrated geometries of airframe (HWB) and propulsion are accounted for simultaneously in a simple algebraic manner, via parameterization of the planform and airfoils at the design sections of the wing body. This paper presents a design of a 300-passenger aircraft that employs distributed electric fans for the propulsion. The trim condition for stability is achieved through the use of the wing tip twist angle. The geometric shape variables are determined through the adjoint optimization method by minimizing the drag while subjecting them to lift, pitching moment, and geometry constraints. An Euler model-based aerodynamic shape optimization is employed to save the design cost for the evaluation of the static margin and longitudinal stability, while the performance of the optimized configuration is evaluated by the RANS model coupled with a drag decomposition method to assess the true aerodynamic performance. The design results clearly show the influence on the aerodynamic characteristics of the installed nacelle and trimming for stability. A drag minimization with the trim constraint yields a reduction of 10 counts in the drag coefficient from the baseline design N3-X configuration, which is comparable with 2000 lbs more payload on a conventional subsonic civil transport airplane.

Appendix

Boundary-layer ingesting (BLI) propulsion has long been exploited by propulsion devices for marine craft, and significant improvements have been achieved because the flow physics involves the low Reynolds flow regime. Even though its benefit is not expected to be as great as in marine applications due to the much thinner wake generated on air vehicles, interest in its application has arisen in the aviation community. The BLI effect is still under exploration due to few experiments and studies having been conducted, especially from the perspective of propulsion–airframe integration (PAI). After the present PAI configuration aerodynamic study was completed in 2017, the N3-X-DEP300 development has been focused on the design and optimization of its propulsion system, including the components of inlet/nozzle, i.e., the flow path [29], and sets of counter-rotating fans [30]. The authors not only realized a need to address the reasoning why the Euler model-based optimization was employed in the present paper over the RANS-based one, but also felt the desire to share the knowledge and lessons learned in the BLI propulsion. Meanwhile, the authors would like to keep the original form of the paper as close as possible to what our deceased co-author last saw. In lieu of rewriting the paper, the authors thus leave the following justification and discussion regarding the multidisciplinary subject of PAI as notes in this Appendix.

Choosing the inviscid assumption for the PAI configuration design is for two reasons. Firstly, the optimization results are mainly driven by the reduction in the induced drag. The induced drag of the present hybrid wing body configuration is found to be almost independent of both the wave and viscous drag components as the baseline configuration does not experience significantly strong shock waves or shock-induced separation on the main airframe (see Table 3). Hence, the results from the mid-field drag decomposition method for the original N3-X PAI configuration [27] showed that the lift-to-induced-drag ratio estimated by the Euler analysis agrees well with the RANS model. Secondly, the propulsion modeling that the current study adopted is based on the 1-D model and setting a CFD outlet boundary condition at the inlet face. However, this coupling does not properly work in the RANS model, while it works reasonably in the Euler model for the design mass flow rate condition. In the case of the RANS model, the 1-D back pressure boundary condition in a CFD domain cannot simulate the static pressure profile at the inlet face; thus, the boundary-layer profile could be easily distorted due to numerical issues. Meanwhile, the Euler model can decouple the effect of the static pressure distortion from the shape of the ingested flow stream.

To validate the second argument, the effect of the propulsion models toward the static pressure distortion in the root section (y/b₂ = 1.5%) is compared in Fig. 21 for an optimized N3-X-DEP300 PAI configuration. Both RANS and Euler analyses adopt the propulsor design mass flow rate condition, $\dot{m}$ = 810 kg/s (a cruise flight condition). Note that the propulsion models, also referred to as engine models, applied in this study are the 1-D boundary condition with NPSS data [9], an actuator disk (AD) model [31], and a body force model [13, 30] proposed for a BLI propulsor. The last two are higher-fidelity propulsion models than the 1-D model. Figure 21b, d shows the pressure contours of the AD and body force models, respectively, and indicates two series of shock waves on the nacelle outer mold line (OML). The inviscid calculation seems to capture the same feature as shown in Fig. 21a, while the RANS result with the 1-D model in Fig. 21c fails to predict the first shock wave. Here, it is noted that all the RANS calculations are done on the same level of the mesh refinements with identical surface mesh distribution while the inviscid calculation is directly from a coarser mesh that is used during the Euler model-based optimization.

Fig. 21

Comparison of the flow field around the propulsion system per the propulsion model (y/b₂ = 1.5% (inboard), M_∞ = 0.84, Re/m = 7.2 × 10⁶, AOA = 3.06°, $\dot{m}$ = 810 kg/s, 40,000 ft altitude ISA)

Figure 22 depicts flow features further close-up near the inlet region of all the cases. A noticeable difference in the exit pressure contour level at the upstream of the inlet is observed between the inviscid (Fig. 22a) and RANS (Fig. 22c) results, both with the 1-D boundary condition, i.e., mass flow outlet boundary condition. Considering the flow blockage due to the low-momentum flow near the airframe wall in the boundary-layer ingestion cases (RANS cases), the boundary condition tends to lower the back pressure to redeem the desired mass flow rate. The lowered constant back pressure, which is lower than those of the high-fidelity engine models in Fig. 22b, d at the shroud region, generates an excessive suction effect; thus, a strong shock accompanied with a significant expansion is observed near the inlet throat region in Fig. 22c. Hence, the incidence angle of the ingested stream tubes near the inlet highlight differs from the cases taking the static pressure distortion at the inlet face into account, as shown in Fig. 22b, d. In the case of the inviscid assumption in Fig. 22a, the trend is opposite. The flow near the airframe wall still keeps momentum; the ingested stream tube near the wall thus diffuses. The effect generates positive incidence toward the inlet highlight. Consequently, the static pressure distribution near the airframe wall in Fig. 22a is observed higher than in other cases, but the pressure distribution and incidence near the shroud and inlet highlight are kept similar to those of the high-fidelity engine models.

Fig. 22

Comparison of the flow field around the propulsion system per the propulsion model near inlet (y/b₂ = 1.5% (inboard), M_∞ = 0.84, Re/m = 7.2 × 10⁶, AOA = 3.06°, $\dot{m}$ = 810 kg/s, 40,000 ft altitude ISA)

Figure 23 displaying the surface pressure distribution along the section at y/b₂ = 1.5% near the wing root area shows the difference in a quantitative manner. The different mechanism of the ingested stream tube shaping caused by the 1-D boundary condition affects the upstream potential flow field as shown in Fig. 23. Up to x = 17 m, the pressure distributions show a good agreement, while the cases start showing deviation from each other as the flow stream tube encounters an adverse pressure gradient. Interestingly, the RANS result with the 1-D engine model that has the boundary-layer profile is affected much more than the inviscid result. The fact that the potential flow field on the airframe could be affected by the engine model is indicated by the AD model showing a small shock structure at x = 21 m, while others do not. The inviscid model follows well with the trend of the two high-fidelity models even at the near trailing edge and the nacelle OML regions, while the RANS model with the 1-D boundary condition shows a significant deviation. The only deviation that the inviscid model exhibits could be found at the upper wing surface near the inlet entrance about x = 29–31 m due to the above-explained stream tube shaping at the airframe wall.

Fig. 23

Normalized surface pressure distribution (P_s/P_s,∞) comparison along the flight direction (m) (y/b₂ = 1.5% (inboard), M_∞ = 0.84, Re/m = 7.2 × 10⁶, AOA = 3.06°, $\dot{m}$ = 810 kg/s, 40,000 ft altitude ISA)

The bottom lines derived from the authors’ studies are as follows:

• 1. The propulsion system affects the fuselage potential flow field. The authors would not name it a “BLI effect,” but “the effect of the embedded propulsion system toward the fuselage aerodynamics.”

• 2. However, the effect is limited to the location where the adverse pressure gradient starts along the flight direction, i.e., at x = 17 m location in Fig. 23b, in high transonic flight conditions.

• 3. The effect of the propulsor operation cannot be appropriately incorporated without knowing the propulsor design. For instance, the body force model can only be incorporated after the conceptual propulsor design [30] is completed and the AD model needs at least the flow path geometry inside the nacelle. Hence, from the perspective of PAI, the impact of the employed propulsor model is more significant than the fidelity of the flow analysis methods.

• 4. The actuator disk or zone models predict much better than the 1-D model does in terms of the modeling of the static pressure distortion at the inlet face of the embedded system. The impact toward the potential flow field over the upstream airframe is more critical than the debate about the Euler and RANS models.

• 5. With the 1-D boundary condition, the RANS model is not able to predict the impact of the propulsor appropriately for the design mass flow rate condition as presented in Figs. 21, 22, and 23, while the Euler model provides a better interpretation of the airframe performance which is decoupled from it. Thus, the Euler model seems more appropriate in the case that the airframe designer does not know much about the effect of the engine operation.

• 6. The aft-loaded shock location and shock-induced separation could have misled the Euler model-driven aero-design, but the present study of the N3-X-DEP300 is not the case. Tables 2 and 3 show that the drag caused by the shock loss in the optimized configuration is less than 1% of the total drag for both the Euler and RANS models. In addition, no change in the viscous loss between the two models is indicated in Table 3. Thus, the Euler model-based optimization adopted is appropriate in the present study, but by no means can be generalized for other cases.