Parametric Analysis of Different Nacelle Positions in the

DLR-F6 Model by Means of the CFD++ Code

Ana Marta de Souza1 and Aristeu da Silveira Neto2 Federal University of Uberlandia, Uberlandia, Minas Gerais, Brazil, 38400-902

Francisco José de Souza3, Antonio Batista de Jesus4 and Guilherme Lara Oliveira5 EMBRAER – Empresa Brasileira de Aeronáutica, São José dos Campos, São Paulo, Brazil, 12227-901

and

João Luiz F. Azevedo6 Instituto de Aeronáutica e Espaço, São José dos Campos, São Paulo, Brazil, 12228-903

The integration of engines into the aircraft involves many important aspects. These

aspects can be investigated by means of wind tunnel experiments or numerical simulations.

The main goal of this study is to analyze the effects of different nacelle positions on the

aerodynamic coefficients and pressure distributions through numerical simulations. The

aircraft geometry investigated is the DLR-F6 configuration. The nacelle is changed from the

original configuration to four different positions, which are characterized by distances in the

longitudinal and vertical directions. High-quality hexahedral grids are used in the CFD

calculations. In order to evaluate the effects of the nacelle shift, all simulations are carried

out in the same conditions and the drag, lift and pressure coefficients are evaluated

numerically. Interesting results are obtained. The lift coefficient is shown to be more

sensitive to variations of nacelle position than the drag coefficient. Based on the simulation

results, it can be inferred that the optimum position for the nacelle, among those

investigated, is the most forward one.

I. Introduction

CCORDING to Harris1, engine/airframe integration can be defined as “that process which is used whenever the performance of the integrated engine/airframe, when operated in a designed combination, is significantly

different from the sum of the individual engine and airframe performances, that is, for given values of flight Mach number, angle of attack and power setting”. Nicholson’s2 argument, which holds today, was that “the aircraft cannot be conceived first and the propulsive units considered afterwards”. To a very large degree, it can be argued that the engine airframe integration process is at the heart of the overall design process. The focus in design of an aircraft must not rest too long on the individual components or the integration process will be entered too late and, then, the production can get very costly in time, manpower and in terms of lack of performance achievement.

Computational Fluid Dynamics (CFD) has been extensively used in the aircraft industry to evaluate aerodynamic performance during the conceptual and preliminary design stages. With the recent advances in CFD and computer capabilities, it is now possible to simulate complete airplane configurations in a short enough period of time and to have a significant impact on the design cycle. It is generally agreed that CFD is a valuable tool for evaluating the rate of change in the aerodynamic characteristics, such as drag or lift, due to design alterations. Nevertheless, there is still significant uncertainty about the accuracy of CFD for predicting the absolute value of the aerodynamic 1 Research Scientist, School of Mechanical Engineering, M Building, Av. João Naves de Ávila, 2160. 2 Associate Professor, School of Mechanical Engineering, M Building, Av. João Naves de Ávila, 2160. 3 Propulsion Systems Engineer, Av. Brigadeiro Faria Lima, 2170. 4 Propulsion Systems Engineer, Av. Brigadeiro Faria Lima, 2170. 5 Senior Scientist, Av. Brigadeiro Faria Lima, 2170. 6 Senior Research Engineer, Aerodynamics Division, CTA/IAE/ALA, Associate Fellow AIAA. E-mail: azevedo@iae.cta.br.

A

26th AIAA Applied Aerodynamics Conference18 - 21 August 2008, Honolulu, Hawaii

AIAA 2008-7053

Copyright © 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.

characteristics, particularly the drag coefficient. As a result of this lack of confidence, the current state of the art is to use CFD as a tool for screening a large number of potential designs. The best candidates are, then, selected for further testing in a wind tunnel where the actual values of the aerodynamic characteristics are measured and used for all the performance calculations.

The main goal of this work is to evaluate the nacelle position in an aircraft with wing-mounted nacelles without thrust, by means of pressure distributions, lift and drag coefficients. The baseline case chosen is the wing-body-nacelle-pylon (WBNP) DLR-F6 configuration with through-flow nacelles (TFN). This is a public domain geometry for which there is a large body of high-quality experimental data available. 3

II. Baseline Case

Figure 1 illustrates the WBNP DLR-F6 model (Brodersen4) with engines mounted under the wings, which was one of the configurations simulated in the present work. This configuration is relatively complex and represents a typical twin-engine wide body aircraft.

Figure 1. DLR-F6 geometry with underwing–mounted engines (dimensions in mm) (Brodersen4).

The mean aerodynamic chord of the DLR-F6 is 0.1412 m, and its half-model reference area is 0.0727 m2. The

nacelle centerline is defined by the following two points: Point 1: x=-8.73756 mm, y=-205.39937 mm, z=-28.54676 mm. Point 2: x= 203.95542 mm, y= -209.07073 mm, z= -35.98394 mm. The characterization of the nacelle positions investigated is presented in Fig. 2. Essentially, the parameters employed to describe each position are the distances in the longitudinal and vertical directions between the wing leading edge and the nacelle upper trailing edge, XTE and ZTE, nondimensionalized by the local wing chord length, c.

Figure 2. Definitions of nacelle position (Brodersen4).

The shift of the nacelle in the XTE and ZTE directions, in relation to the original DLR F-6 configuration,

determined new nacelle positions, which are presented in Table 1. Figure 3 shows the baseline and the other four nacelle positions studied in the present work.

Table 1 - Nacelle positions investigated.

Nacelle Position XTE /c ZTE /c Baseline 0.4895 -0.2026

x1 0.3000 -0.2026 x2 0.2000 -0.100 z1 0.4895 -0.2679 z2 0.3000 -0.100

(a) Baseline (XTE/c=0.4895 and ZTE /c =-0.2026)

(b) x1(XTE/c=0.3000 and ZTE /c =-0.2026)

(c) x2(XTE/c=0.2000 and ZTE /c =-0.100)

XTE

ZTE

(d) z1(XTE/c=0.4895 and ZTE /c =-0.2679)

(e) z2(XTE/c=0.300 and ZTE /c =-0.100)

Figure 3. Visualization of the nacelle positions.

III. Setup of the Simulations

A. Grids

Since the main interest of this investigation is in the cruise condition, only half-span models are considered. The grids employed contain nearly 6 million hexahedra. The grids are created with the mesh generator ICEM 5.15. The quality of the meshes is rather satisfactory, with most of the elements with determinants above 0.6.

B. Solver Setup

In order to compute the flow around the five configurations, the CFD++6 commercial code is used. The freestream boundary conditions are set as follows:

Static Temperature: T∞ = 305 K; Static Pressure: P∞ = 132000 N/m2; Velocity: V = 262.7 m/s; Turbulence intensity: 0.1 %; Turbulence viscosity/molecular viscosity ratio: 1.

In order to perform the parametric analysis of the DLR-F6 configurations, the angle of attack α = 1.0o is maintained in all simulations. It is important to mention that, in a real flight scenario, one is typically concerned with achieving a pre-defined lift coefficient by changing the angle of attack. Therefore, it should kept in mind that, perhaps, a more realistic way of running the present investigation would be to keep the lift coefficient constant for the 5 geometries instead of the angle of attack. Nevertheless, the results are still significant in clarifying the effects of geometry variations.

At the temperature and pressure of the experiments, the density and molecular dynamic viscosity of the air are,

respectively, 1.506 kg/m3 and 1.87 x 10-5 kg/(m.s), which yield an approximate Reynolds number of Re = 3 million. The turbulence effects are accounted for by the realizable k-ε model6, and the advanced wall functions available in CFD++6 are used on all solid surfaces. Freestream boundary conditions are modeled based on characteristics.

IV. Results

A. Parametric Analysis of Nacelle Positions on DLR-F6

Table 2 compares the drag and lift coefficients of the different DLR-F6 configurations. The comparison is

presented in terms of the coefficient deviation (∆) between the new nacelle position and the baseline configuration. Moreover, the lift-to-drag ratio (L/D) is also provided for all cases.

Table 2 – Drag and lift coefficients for the different DLR-F6 configurations.

DLR-F6 WBNP TFN

Configuration Cd ∆Cd (%) Cl ∆Cl (%) L/D

Baseline 0.03798 - 0.53272 - 14.026 x1 nacelle position 0.03810 0.32 0.55225 3.67 14.495 x2 nacelle position 0.03841 1.13 0.56236 5.56 14.641 z1 nacelle position 0.03789 -0.24 0.52595 -1.27 13.881 z2 nacelle position 0.03826 0.74 0.55765 4.68 14.575

The pressure coefficients at various spanwise locations along the wing, for the different nacelle positions, are shown in Fig. 4. It can be seen from the second inboard spanwise plot that the original DLR-F6 presents a strong nacelle-pylon-wing interference, with the flow excessively accelerating in the channel between the wing lower surface and the nacelle. This effect persists with the z1 position, which means that simply displacing the nacelle vertically downwards does not mitigate this interference. However, unlike the baseline and z1 positions, the other 3 positions apparently attenuated this virtual channel effect, decreasing the flow velocity between the nacelle and the wing. An analysis of all four spanwise pressure coefficient plots allows the conclusion that the largest gains in decelerating the flow in the virtual channel region, and therefore increasing pressure, are obtained for the x2 nacelle position. As a matter of fact, the results in Table 2 reinforce such a conclusion, by showing that the highest lift-to-drag ratio is observed for the x2 position. Even though the pressure coefficient plots also indicate an increase of Cp along the wing upper surface, the overall result for the x2 position is a gain in lift.

Figure 4. Pressure coefficients computed at four different wing span locations for the different nacelle

positions.

Figure 5 shows the pressure coefficient distributions on three different nacelle azimuthal sections (60o, 180o and 300o) for the different nacelle positions. These Cp plots confirm that the most important effects of the nacelle positioning occur inboard, close to the pylon. The 60o plot also confirms the similarity in the Cp distributions between the original DLR-F6 and z1 positions, which display strong virtual channel effects. In analyzing the results of the simulations of the present work, it is important to bear in mind that, since the nacelles consider a flow-through condition, there is no control over the mass flow rates across them. Such a condition is in contrast with real engines, for which the mass flow rate should remain the same in all nacelle positions. Nevertheless, since the nacelle displacements are relatively small, it is not expected that the possible mass flow rate difference compromises the conclusions herein. Indeed, the nacelle pressure coefficient plots indicate that the stagnation point remains parametrically the same for all nacelle positions.

The results showed that the lift coefficient is more sensitive to nacelle shifts. Forward shifts (x1, x2 and z2 nacelle positions) lead to increases of the lift coefficient. The downward shift of the nacelle (z1 position) causes a decrease in the lift coefficient. Among all the positions investigated, the x2 nacelle position seems to be the “optimal” nacelle position, as it leads to the highest gain in the lift coefficient, without causing significant increase in the drag. Furthermore, the pressure coefficients over the wing show that the x2 nacelle position tends to minimize the underwing shock waves that appear in other nacelle positions, as well as in the baseline case (see Fig. 4).

Figure 5. Measured and computed pressure coefficient on nacelle sections (60o, 180

o and 300

o) for the five

cases investigated.

Figure 6 is a classic guide to nacelle positioning. It displays a bell-shaped region under which interference drag

is unacceptable. The relative positions investigated in this work are also plotted in the figure. It can be seen that the z1 position is the one that lies furthest from the interference region. By analyzing Table 2 and Fig. 6, a clear tendency for increasing lift-to-drag ratio is noticed as one moves towards the bell region. Indeed, as concluded

above, the x2 location appears to be the best position, whereas z1 is the worst one in terms of L/D. Positioning the nacelle within the bell region and analyzing the aerodynamic effects are the object of future investigation.

Figure 6. Interference region as a function of the relative nacelle position.

B. Flow Visualizations Figure 7 displays the experimental oil-flow visualization and CFD predictions of the wing-engine-pylon

separation. This reverse flow region is a known effect of the nacelle-pylon installation and it can be easily seen in the experimental visualization. However, it is not possible to observe the reverse flow through the shear lines presented in Figs. 7(b) to 7(e). At a first glance, one might argue that the failure of the simulation in capturing such feature can be attributed to insufficient mesh resolution. However, at least in the wall-normal direction, this may not be the case, since the y+ values in the first cells away from the wall are nearly 1.

(a) experimental

(b) baseline

baseline

z1 x1

z2 x2

(c) x1 nacelle position

(d) x2 nacelle position

(e) z1 nacelle position

(f) z2 nacelle position

Figure 7. Experimental visualization of recirculation region and shear lines for the five configurations.

Figure 8 presents the pressure distributions over the wing lower surface for the five nacelle positions. The nacelle is not shown in the figures. The pressure decrease is smaller for the x2 and z2 nacelle positions.

(a) baseline

(b) x1

(c) x2

(d) z1

(e) z2

Figure 8. Pressure distributions over the lower wing for the different nacelle positions.

V. Conclusions

The paper presents a study of the influence of the nacelle position on the lift and drag coefficients for configurations based on the DLR-F6 wing-body-nacelle-pylon model. Pressure coefficient distributions and flow visualization figures also help identify the relative merits of the different nacelle positions. The results obtained indicate that the lift coefficient is more sensitive to variations of nacelle placement than the drag coefficient. The shift of the nacelle from the baseline case to the x2 position leads to a considerable gain in lift, without causing an equivalent increase in drag. Hence, the results are indicating that a forward and upward shift of the nacelle seems to yield the best results among the test cases investigated.

The computational results have not been able to detect the reverse flow region which is present at the wing-pylon intersection in the experimental investigation. It is not believed that such result can be directly attributed to insufficient mesh resolution, as the y+ values on the cells away from the wall are nearly 1. In any event, grid refinement is one aspect that will be investigated in future work in an attempt to clarify and further understand the present results.

Acknowledgements

The authors acknowledge the support of Fundação de Amparo à Pesquisa do Estado de São Paulo, FAPESP, through Research Grant No. 2000/13768-4. The last author also acknowledges the partial support of Conselho Nacional de Desenvolvimento Científico e Tecnológico, CNPq, through the Research Grant No. 312064/2006-3.

References 1Harris, A.E., “Test Techniques for Engine/Airframe Integration,” AGARD 69th Fluid Dynamics Panel Meeting

and Symposium on Aerodynamic Engine/Airframe for High Performance Aircraft and Missiles, Fort Worth, TX, Sept. 1992.

2Nicholson, L.F., “Engine-Airframe Integration,” RAeS Lecture, April 1957 (published in RAeS Journal, Vol.

61, Nov. 1957). 32nd DPW, 2nd Drag Prediction Workshop, http://ad-www.larc.nasa.gov/tsab/cfdlarc/aiaa-dpw/, 2003.

4Brodersen, O., “Drag Prediction of Engine-Airframe Interference Effects Using Unstructured Navier-Stokes Calculations,” Journal of Aircraft, Vol. 39, No. 6, Nov.-Dec. 2002, pp. 927-935.

5ANSYS ICEM CFD, Software Package, version 5.1, ANSYS, Inc., Berkeley, CA, USA. 6Metacomp Inc., CFD++, version 5.2, Metacomp Inc, Agoura Hills, CA., USA. 7Rudnik, R., Rossow, C.-C., and von Geyr, H.F., “Numerical Simulation of Engine/Airframe for High Bypass

Engines,” Aerospace Science and Technology, Vol. 6, No. 1, Feb. 2002, pp. 31-42.