Académique Documents
Professionnel Documents
Culture Documents
Results from the Second AIAA Drag Prediction Workshop are summarized. The workshop focused on absolute
and configuration delta drag prediction of the DLR, German Aerospace Research Center F6 geometry, which is
representative of transport aircraft designed for transonic flight. Both wing–body and wing–body–nacelle–pylon
configurations are considered. Comparisons are made using industry relevant test cases that include single-point
conditions, drag polars, and drag-rise curves. Drag, lift, and pitching moment predictions from several different
Reynolds averaged Navier–Stokes computational fluid dynamics codes are presented and compared to experimental
data. Solutions on multiblock structured, unstructured, and overset structured grids using a variety of turbulence
models are considered. Results of a grid-refinement study and a comparison of tripped transition vs fully turbulent
boundary-layer computations are reported.
Center- (DLR-) F4 wing–body configuration was chosen as the sub- These comments guided the organizing committee to choose the
ject of DPW-I because of both its simplicity and the availability of DLR-F6 wing–body (WB) and wing–body–nacelle–pylon (WBNP)
publicly released experimental test data.2 The workshop commit- configurations. Test cases were chosen to focus on absolute drag and
tee provided a standard set of multiblock structured, unstructured, component drag increment prediction accuracy. Each participant
and overset structured grids for the DLR-F4 geometry to encour- was asked to perform grid-refinement studies of both configurations
age participation in the workshop and reduce variability in the CFD at a fixed-C L single-point condition and to submit computed drag
results. However, participants were encouraged to construct their polars for both configurations. Additionally, participants were en-
own best practices grids, so that learned knowledge concerning grid couraged to submit computed drag-rise curves and perform specified
generation and drag prediction might be shared among the work- tripped transition and fully turbulent boundary-layer computations
shop attendees. The test cases were chosen to reflect the interests for both configurations.
of industry and included a fixed-C L single-point solution, drag po- DPW-II was sponsored by the AIAA Applied Aerodynamics
lar, and constant-C L drag-rise data sets. There were 18 participants Technical Committee and held 21–22 June 2003 in conjunction
who submitted results using 14 different CFD codes; many submit- with the 21st AIAA Applied Aerodynamics Conference in Orlando,
ted multiple sets of data exercising different options in their codes, Florida. The DPW series has spawned considerable interest in the
for example, turbulence models and/or different grids. Because of CFD and applied aerodynamic communities, and this summary pa-
strong participation, DPW-I successfully amassed a CFD data set per is one of several DPW-II papers.9−23 Approximately 75 persons
suitable for statistical analysis.3 However, the results of that analysis from five continents representing academia, research laboratories,
were disappointing, showing, for example, a 270 drag count, that and industry attended the workshop. Of those attending, 25 partici-
is, drag coefficient unit = 0.0001, spread in the fixed-C L data, with pated in the workshop, submitting data for analysis. Those submit-
a 100:1 confidence interval of more than ±50 drag counts. ting data were given the opportunity to present their results. The
Despite the disheartening results of the statistical analysis, DPW-I workshop data and presentations may be viewed at the DPW-II
was a definitive success. It brought together CFD developers and website.24
practitioners and focused their efforts on a common problem. It fa- In the following sections, a description of the DLR-F6 geom-
cilitated an exchange of learned best practices and promoted open etry and the wind-tunnel experimental approach is included. The
discussions, identifying areas requiring further research or addi- standard multiblock structured, unstructured, and overset structured
tional scrutiny. Possibly most significantly, it employed statistical grids, provided to participants by the DPW-II organizing committee,
methods to assess CFD results objectively. Finally, it reminded the are briefly described. Test cases are defined, and an overview of the
CFD and applied aerodynamics communities that CFD is not yet a participation and submitted data is presented. The DPW-II results
fully mature discipline. are summarized for each of the four test cases, which include a grid-
In addition to the accomplishments just listed, DPW-I created a refinement study and configuration delta drag comparisons. Lift,
sustained interest in industry relevant drag prediction. Several of the drag, and pitching moment characteristics for both the DLR-F6 WB
participants presented their DPW-I results at a well-attended special and WBNP configurations are discussed. Drag data are compared
session of the 2002 AIAA Aerospace Sciences Meeting and Exhibit against grid size, grid type, turbulence model, and boundary-layer
in Reno, Nevada.3−8 transition specification. Finally, conclusions of DPW-II and recom-
The interest generated by DPW-I naturally led to the planning mendations for future drag prediction workshops are presented.
and organization of a Second AIAA Drag Prediction Workshop
(DPW-II). The DPW-II organizing committee, recognizing the suc- Geometry Description
cess of DPW-I, maintained the DPW-I objectives for DPW-II. These The configuration selected for DPW-II is a wing-mounted twin-
objectives are as follows. engine aircraft representing a typical modern wide-body transport.
1) State-of-the-art computational methods are to be assessed as This configuration, called DLR-F6, was designed by DLR more
practical aerodynamic tools for aircraft force and moment prediction than 20 years ago and is derived from the DLR-F4 configuration,25
of industry relevant geometries, focusing on drag prediction. which was the subject of DPW-I (Ref. 3). The DLR-F6 fuselage
2) An impartial forum is to be provided for evaluating the effec- remained unchanged from the DLR-F4 configuration. However, the
tiveness of existing computer codes and modeling techniques using outer three of the four wing defining airfoils were modified. The
Navier–Stokes solvers. airfoil at the wing kink is translated upward to achieve a smoother
3) Areas needing additional research and development are to be upper wing surface. The twist distribution is changed, and the span-
openly discussed and identified. Furthermore, the DPW-I format wise position of section three was moved from η = 0.7 to η = 0.84.
was retained, consisting of following key elements. The intention of these changes was to achieve a more elliptic lift
The first element is a common subject geometry, having publicly distribution to reduce or eliminate boundary-layer separation at the
available experimental data, that is simple enough to do high-quality rear upper wing surface of the DLR-F6 wing as compared to the
computations and that is relevant to industry. DLR-F4 wing; however, the goal of eliminating the trailing-edge
The second element consists of required and optional industry separation was not achieved. A list of the DLR-F6 quantities is
relevant test cases. These included a fixed-C L single-point condition, given in Table 1.
drag polars, and constant-C L drag-rise curves. Between 1990 and 1998 several measurement campaigns were
The third element is a standard set of provided grids to encourage carried out in the ONERA-S2MA wind tunnel to investigate the
participants and reduce variability in the CFD results. influence of different flow through nacelles on the DLR-F6 wing
The fourth key element is a rigorous statistical analysis of the flowfield. A model of 1.1713-m span was manufactured by DLR for
CFD results to establish confidence levels in the data.
The fifth element consists of scheduled open forum sessions Table 1 DLR-F6 quantities
to encourage discussion and interaction among participants and
attendees. Property Value
The DPW-II organizing committee relied heavily on discussions S/2 72,700.0 mm2
that took place during DPW-I in identifying what geometry and test c 141.2 mm
cases to use for DPW-II. These discussions focused on poor grid b/2 585.647 mm
quality/resolution and inconsistent boundary-layer transition spec- AR 9.5
ification, that is, tripped transition vs fully turbulent calculations, X ref 157.9 mm
as possible causes for the large scatter in the DPW-I results. Addi- Yref 0.0 mm
Z ref −33.92 mm
tionally, many attendees believed that a second workshop should
Mdes 0.75
focus on a more complex model. It was also hypothesized that C L des 0.5
CFD better predicts configuration delta drag than absolute drag Rec 3 × 106
values.
LAFLIN ET AL. 1167
Standard Grids
To minimize variation in the results and encourage participation,
Fig. 1 DLR-F6 model in ONERA S2MA wind tunnel. a set of standard grids was generated for both the DLR-F6 WB
1168 LAFLIN ET AL.
Table 2 Medium density grid construction specifications Table 3 Multiblock structured grid sizes
Table 6 Overset grid sizes (total nodes, compare nonblanked nodes) The drag polar, case 2, is typical of how industry collects and ex-
amines wind-tunnel experimental data. Participants were requested
DLR-F6 NWB × 10−6 NWBNP × 10−6 to trip the wing boundary layer, that is, specify the position where
Coarse 1.9 3.1 the boundary layer is forced to transition from laminar to turbulent,
Medium 6.8 10.7 at experimental trip locations. Fully turbulent computations were
Fine 23.1 35.8 accepted if a trip position could not be specified using a partic-
ular Navier–Stokes code. Case 2 results were requested to assess
it is likely that shocks occur. An attempt was made to achieve a the effects of grid type, turbulence model, boundary-layer transi-
smooth transition between the prismatic and tetrahedral elements. tion specification, and grid size on computed lift, drag, and pitching
The solution-based grid adaptation of TAU, using pressure and moment.
velocity gradients, has been applied successfully in the past for Optional case 3 was requested to compare the difference in con-
these kinds of configurations and flow conditions and, therefore, figuration delta drags when simpler fully turbulent boundary-layer
was used here to generate the three grid resolutions for each specification is used instead of the more physically correct tripped
configuration.36−38 The grid densities are listed in Table 5. transition boundary-layer specification.
Constant-C L drag-rise curves, case 4, are often used by industry
Overset Structured Grids to examine experimental results and make configuration decisions.
The overset structured grids for the DLR-F6 WB and WBNP con- This case was chosen to compare predicted absolute drag and con-
figurations were developed per specifications defined by the DPW-II figuration delta drag-rise curves to experimental data.
organizing committee. The first specification stated that the medium
grids should be representative of (or slightly more dense than) those Overview of Methods and Data Submitted
currently being used within industry for engineering-level accurate, Data from 22 different Navier–Stokes codes were submitted by
CFD-based drag predictions. To accomplish this, the Boeing Phan- 25 participants. Many participants submitted more than one set of
tom Works team created a 12-zone grid system for the WB configu- results, exercising different options in their codes, for example, tur-
ration that consisted of about 6.8 million grid nodes, then augmented bulence models and/or using different grids. All submitted data sets
this system with an additional 11 blocks to create a 23-zone grid can be found in their entirety at the DPW website.24
system for the WBNP configuration comprising nearly 10.7 million At the conclusion of DPW-II, participants were asked to complete,
grid nodes. A detailed description of these overset grids is provided recheck, and resubmit their data for inclusion in this summary pa-
in Ref. 39. The corresponding coarse/fine overset grid systems were per. This request was made to give participants the opportunity to
derived from the medium grids by decreasing/increasing each grid complete computations and correct errors discovered in their data
dimension by a factor of 1.5 in each zone. In this manner, the size entries. Regrettably, five participants withdrew their data submittals
of the grids in the coarse–medium–fine sequence grows by a factor from consideration or did not resubmit final data sets and so their
of about 3.375. This parametric variation of grid size conforms to results are not represented in this paper.
the DPW-II specifications. Table 6 provides the grid sizes. Tables 7 and 8 summarize the 29 data sets that were resubmitted
for analysis. Table 7 lists the organization submitting the data set,
Test Case Description the code used, and which cases were included in the set. The list
Participants were asked to submit data for two required test cases, is not in any particular order. Both complete and partially complete
case 1 and case 2, and were encouraged to complete two optional data sets are included. Additionally, a solution index i is assigned
test cases, case 3 and case 4 (WB and WBNP, respectively). Be- to each data set, so that the information in Table 7 can be cross
cause each test case involved computations for the WB and WBNP referenced with the information in Table 8.
configurations and either multiple grids or multiple test conditions, Table 7 Participants and submitted data
completion of all four test cases required a significant computational
effort. i Organization Code Cases
The required cases are described next. Case 1 is a fixed-C L 1 Airbus Industries ELSA 1, 2, 3
single-point grid refinement study (six solutions in total), where 2 CFX, Ansys CFX-5 1, 2
M∞ = 0.75 and C L = 0.500 ± 0.001. The boundary layer is fully 3 Boeing Commercial CFL3D 1, 2
turbulent. Coarse, medium, and fine grids are used. 4 Boeing Commercial CFL3D 1, 2, 3, 4
Case 2 is a drag polar study (14 solutions in total), where 5 Boeing Commercial TLNS3D 2
M∞ = 0.75 and α = −3, −2, −1.5, −1, 0, 1, and 1.5 deg. There is 6 Boeing Phantom Works OVERFLOW 1, 2, 3
boundary-layer transition at the lower surface quarter chord and up- 7 Cessna NSU3D 1, 2,
8 Cobalt Solutions COBALT 1, 2
per surface at 5% chord at root, 15% chord at kink, 15% chord at
9 DLR TAU 1, 2, 3, 4
η = 0.844, 5% chord at tip, or 10% chord if trip location cannot be 10 ONERA TAU 2
varied, or the boundary layer is fully turbulent if the trip location 11 Swedish Defence Agency EDGE 1, 2
cannot be specified. A medium grid from case 1 or the participants’ 12 Metacomp Technologies CFD++ 1, 2
own best practices grid is used. 13 Korean Advanced Institute KFLOW3D 1, 2, 4
A description of the optional cases is given next. of Science and Technology
Case 3 is a comparison of tripped and fully turbulent solutions 14 Kawasaki Heavy Industries UG3 1, 2, 3
(four solutions in total), where M∞ = 0.75 and C L = 0.500 ± 0.001. 15 National Aeronautical Laboratory UPACS 1, 2, 3
There is a fully turbulent boundary layer and the boundary-layer 16 NASA Langley Research Center USM3Dns 1, 2
17 NASA Langley Research Center FUN3D 1, 2, 3
tripped transition location as per case 2. The medium grid from
18 NASA Langley Research Center CFL3D 1, 2
case 1 or the participants’ own best practices grid is used. 19 NASA Langley Research Center CFL3D 1, 2, 3
Case 4 is a study of constant-C L drag rise (16 solutions in to- 20 NASA Langley Research Center CFL3D 1, 2
tal), where M∞ = 0.50, 0.60, 0.70, 0.72, 0.74, 0.75, 0.76, and 0.77 21 NASA Langley Research Center CFL3D 1, 2, 3
and C L = 0.500 ± 0.001. There is boundary-layer transition as per 22 NASA Langley Research Center OVERFLOW 1, 2, 3
case 2. The medium grid from case 1 or the participants’ own best 23 NASA Langley Research Center OVERFLOW 1
practices grid is used. 24 National Institute of Aerospace (NIA) NSU3D 1, 2, 3, 4
Participants were asked to perform case 1 to examine the effect of 25 National Aerospace Laboratory ENSOLV 1, 2, 3
grid refinement on absolute and delta drag predictions. In the interest 26 QinetiQ SAUNA 1, 2
27 NIA NSU3D 2
of improving the statistical basis of the data, it was requested that all
28 Science Applications International FEFLO 1, 2
calculations for case 1 be made using fully turbulent boundary-layer Corporation
specifications. Additionally, it was requested that the grids supplied 29 Fluent, Inc. FLUENT 6.1 1, 2
by the DPW-II organizing committee be used if possible.
1170 LAFLIN ET AL.
Table 9 WB and WBNP medium grid sizes Table 10 WB and WBNP coarse and fine grid sizes
Table 13 Case 1, interpolated experimental values Table 14 Case 1: WB coarse grid results
Table 20 Case 1: configuration ∆CD results of 9, 6, and 8 drag counts for the coarse, medium, and fine grid
solutions, respectively. This result indicates that, even with the ex-
Grid Average Minimum Maximum Expmt
istence of significant regions of separation, configuration delta drag
Coarse 0.00520 0.00410 0.00602 0.00090 predictions are largely insensitive to grid density, unlike absolute
Medium 0.00458 0.00342 0.00633 0.00063 drag predictions.
Fine 0.00498 0.00318 0.00600 0.00078
Case 2
Case 2 (required) is representative of a typical constant Mach
number sweep, that is, drag polar, performed in wind-tunnel testing.
Summary results for case 2 include drag coefficient, lift coefficient,
and pitching moment coefficient predictions. Also, drag predictions
are compared against grid size, grid type, turbulence model, and
boundary-layer transition specification to examine how each affect
drag prediction. The reader is reminded that Tables 10 and 11 should
be used as legends for the case 2 plotted results. In the summary
plots of case 2, (see Figs. 12–20) the drag values are presented in
terms of the idealized profile drag coefficient41
a) WB C D P = C D − C L2 /(πAR)
Fig. 9 Delta (WBNP − WB) drag results for case 1: M∞ = 0.75 and
CL = 0.5.
a) WB b) WBNP
Fig. 11 Composite pitching moment results for case 2: M∞ = 0.75.
a) WB b) WBNP
Fig. 12 Composite drag polar results for case 2: M∞ = 0.75.
Fig. 14 WBNP, drag polar trends with grid size for case 2: M∞ = 0.75. Fig. 16 WBNP, drag polar trends with grid type for case 2: M∞ = 0.75.
Fig. 17 WB, drag polar trends with turbulence model for case 2:
Fig. 15 WB, drag polar trends with grid type for case 2: M∞ = 0.75. M∞ = 0.75.
configurations, respectively. As expected, specifying a boundary- Case 3
layer trip position, allowing a run of laminar flow, reduced the Case 3 (optional) is a single fixed-C L and Mach number con-
drag. Additionally, tripping the boundary layer generally reduced the dition using FT boundary-layer specification and fixed position
range in the drag data compared to the FT solutions. It is curious that, transition, that is, tripped boundary layer, specification on both
although the experimental wind-tunnel model used boundary-layer DLR-F6 configurations. This test case was chosen to determine
trips, the FT CFD solutions better predict the experimental results for two different delta drag values. The first delta drag of interest is the
the WB configuration, Fig. 19. However, the tripped CFD solutions difference between FT and tripped boundary-layer solutions. This
better match the WBNP configuration experimental data, Fig. 20. delta drag comparison is summarized in Table 22. Here, the tripped
1176 LAFLIN ET AL.
Fig. 18 WBNP, drag polar trends with turbulence model for case 2:
M∞ = 0.75.
Fig. 21 Delta (FT-Tr) drag results for case 3: M∞ = 0.75 and CL = 0.5.
Conclusions
A workshop was held to assess the state-of-the-art of CFD
Reynolds-averaged Navier–Stokes solvers to predict lift, drag, and
pitching moment of industry relevant aircraft configurations, fo-
cusing principally on drag calculations. Participants were asked to
compute a prescribed set of test cases on the DLR-F6 WB and
WBNP configurations. The test cases consisted of fixed-C L single-
b) WBNP point solutions, drag polars, and constant-C L drag rise curves. Stan-
dardized multiblock structured, unstructured, and overset structured
Fig. 23 Drag rise results for case 4.
grids were made available to workshop participants to encourage
participation and reduce grid variability. Grid refinement, configu-
ration delta drags, and the effect of boundary-layer transition speci-
fication were all examined. A large body of data was gathered from
20 international participants and presented in an objective manner.
In general, the CFD lift levels are higher than the wind-tunnel
results, drag is lower, and pitching moment is more negative (more
nose down) at a given angle of attack for the DLR-F6 WB config-
uration. For the DLR-F6 WBNP configuration, the CFD lift levels
are generally higher, and pitching moment is more negative (more
nose down) than the wind-tunnel results at a given angle of attack.
a) The WBNP configuration CFD-predicted drag is higher then experi-
mentally measured drag for C L values less than about 0.5; otherwise
it is lower.
Although the DPW-II organizing committee envisioned a grid-
convergence study as part of the workshop, the various series of
coarse, medium, and fine grids used by participants were of insuffi-
cient density to obtain asymptotic solution convergence. However,
it was noted that the range in the CFD predictions decreased with
increased grid density, indicating that proper grid resolution may
be the single most important factor in achieving quality CFD drag
predictions, compared to grid type or turbulence model. However,
this conclusion was not substantiated by statistical analysis.
b) In addition to grid size, the effects of boundary-layer transition
Fig. 24 Example of statistical results for CD (WB configuration) from specification, that is, Tr vs FT, turbulence model, and grid type on
Ref. 10. the range in the drag data were examined. None of these appear to
have as significant an effect as grid size. Specifying the boundary-
Case 4 layer trip location appeared to reduce the range in drag as compared
Case 4 (optional) is computed drag rise curves for both DLR-F6 to specifying FT boundary-layer computations. Tripped solutions
configurations. This test case required a significant computational were, on average, about 10 drag counts lower than FT solutions. No
effort because the solution must be converged to a specified C L value conclusive advantage as to grid type or turbulence model could be
for a number of different Mach number conditions. As such, only discerned from the data.
four participants submitted case 4 data; these are shown in Figs. 23a Delta drag comparisons between the DLR-F6 WB and WBNP
and 23b for the WB and WBNPs configurations, respectively. Three configurations were examined. The variation in these two configu-
of the four submitted data sets consistently underpredict drag for rations is more extreme than usually considered in CFD delta drag
both configurations over the range of Mach numbers for which ex- studies. Additionally, CFD drag comparisons that are conducted
perimental data is known. The general shape of the drag rise is for the purpose of making design decisions focus on configuration
captured in the WBNP computations but is generally too flat for the variants absent of large pockets of separated flow, unlike the config-
WB configuration. The tripped solutions show less variance to each urations considered for the workshop. Nevertheless, the workshop
other than do the two FT solutions. The increased variation for the results indicate that Tr boundary-layer treatment is not critical for
FT cases may be due to the unique numerical transition behavior delta drag comparisons. Additionally, CFD delta drag predictions
inherent in all turbulence models, that is, each turbulence model were less dependent on grid density than CFD absolute drag predic-
transitions at a different location even when allowed to run in the tions. These results tend to support the current industry practice of
FT mode. using FT boundary-layer calculations on coarse density grids to ob-
tain configuration delta drag values quickly. This procedure assumes
Statistical Analysis similar errors exist in both CFD solutions and that these errors can-
A rigorous statistical analysis of the DPW-II CFD results was cel each other, resulting in an adequate estimate of delta drag. Ide-
performed and presented during the workshop.24 These results are ally, this procedure should only be performed when configuration
1178 LAFLIN ET AL.
changes are small, flow separation is minimal, and all grids have 15 Wutzler, K., “Aircraft Drag Prediction Using Cobalt,” AIAA Paper
tent on the VGRIDns unstructured grids for this paper. Finally, Complex Three-Dimensional Configurations Employing the DLR TAU-
a special thanks goes to all of the DPW-II participants because Code,” AIAA Paper 97-0167, Jan. 1997.
27 Kroll, N., Rossow, C.-C., Becker, K., and Thiele, F., “The MEGAFLOW
without their contributions this workshop would not have been
possible. Project,” Aerospace Science Technology, Vol. 4, No. 4, 2000, pp. 223–237.
28 Rossow, C.-C., Godard, J.-L., Hoheisel, H., and Schmitt, V., “Investiga-
tions of Propulsion Integration Interference Effects on a Transport Aircraft
References Configuration,” Journal of Aircraft, Vol. 31, No. 5, 1994, pp. 1022–1030.
1 Proceedings of the First AIAA CFD Drag Prediction Work- 29 Pierre, M., and Fasso, G., “The Aerodynamic Test Center of Modane-
shop, URL: http://aaac.larc.nasa.gov/tsab/cfdlarc/aiaa-dpw/Workshop1/ Avrieux,” ONERA TN 166E, March 1972.
workshop1.html, email: dpw@cessna.textron.com [cited 20 Sept. 2004]. 30 Tinoco, E. N., and Bussoletti, J. E., “Minimizing CFD Uncertainty
2 Redeker, G., “DLR-F4 Wing-Body Configuration,” A Selection of Exper- for Commercial Airplane Applications,” AIAA Paper 2003-0407, Jan.
imental Test Cases for the Validation of CFD Codes, AGARD Rept. AR-303, 2003.
Vol. 1, Aug. 1994, pp. 88–89. 31 Capron, W. K., and Smit, K. L., “Advanced Aerodynamic Applications
3 Hemsch, M., “Statistical Analysis of CFD Solutions from the Drag Pre- of an Interactive Geometry and Visualization System,” AIAA Paper 91-0800,
diction Workshop,” AIAA Paper 2002-0842, Jan. 2002. June 1991.
4 Rakowitz, M., Eisfeld, B., Schwamborn, D., and Sutcliffe, M., “Struc- 32 Su, T. Y., Wilkinson, W. M., and Yu, N. J., “Structured Navier–Stokes
tured and Unstructured Computations on the DLR-F4 Wing-Body Configu- Grid Generations for Propulsion-Integrated Airplane Configurations,” 5th
ration,” Journal of Aircraft, Vol. 40, No. 2, 2003, pp. 256–264. International Conference on Numerical Grid Generation in Computational
5 Mavriplis, D. J., and Levy, D. W., “Transonic Drag Prediction Using an Field Simulations, edited by B. K. Soni, J. F. Thompson, J. Häuser, and P.
Unstructured Multigrid Solver,” AIAA Paper 2002-0838, Jan. 2002. Eiseman, NSF Engineering Research Center, Mississippi State, Mississippi,
6 Pirzadeh, S. Z., “Assessment of the Unstructured Grid Software TetrUSS 1996, pp. 1193–1203.
for Drag Prediction of the DLR-F4 Configuration,” AIAA Paper 2002-0839, 33 Lohner, R., and Parikh, P., “Generation of Three Dimensional Unstruc-
Jan. 2002. tured Grids by the Advancing Front Method,” International Journal of Nu-
7 Vassberg, J. C., Buning, P. G., and Rumsey, C. L., “Drag Prediction for merical Methods in Fluids, Vol. 8, No. 10, 1988, pp. 1135–1149.
the DLR-F4 Wing/Body Using OVERFLOW and CFL3D on an Overset 34 Pirzadeh, S., “Three-Dimensional Unstructured Viscous Grids by
Mesh,” AIAA Paper 2002-0840, Jan. 2002. the Advancing Front Method,” AIAA Journal, Vol. 34, No. 1, 1996,
8 Levy, D. W., Zickuhr, T., Vassberg, J., Agrawal, S., Wahls, R., Pirzadeh, pp. 43–49.
S., and Hemsch, M., “Summary of Data from the First AIAA CFD Drag Pre- 35 CentaurSoft, http://www.centaursoft.com [cited 20 Sept. 2004].
diction Workshop,” Journal of Aircraft, Vol. 40, No. 5, 2003, pp. 875–882. 36 Gerhold, T., and Evans, J., “Efficient Computation of 3D-Flows for
9 Laflin, K., Klausmeyer, S., Zickuhr, T., Vassberg, J., Wahls, R., Morrison, Complex Configurations with the DLR-Tau Code Using Automatic Adap-
J., Brodersen, O., Rakowitz, M., Tinoco, E., and Godard, J., “Summary of tation,” Notes on Numerical Fluid Mechanics, edited by W. Nitsche, H.-J.
Data from the Second AIAA CFD Drag Prediction Workshop,” AIAA Paper Heinemann, and R. Hilbig, Vol. 72, Vieweg, Brunswick, Germany, 1998,
2004-0555, Jan. 2004. pp. 178–185.
10 Hemsch, M., and Morrison, J., “Statistical Analysis of CFD Solutions 37 Brodersen, O., “Drag Prediction of Engine-Airframe Interference Ef-
from 2nd Drag Prediction Workshop,” AIAA Paper 2004-0556, Jan. 2004. fects Using Unstructured Navier–Stokes Calculations,” Journal of Aircraft,
11 Brodersen, O., Rakowitz, M., Amant, S., Destarac, D., and Sutcliffe, Vol. 39, No. 6, 2002, pp. 927–935.
M., “Airbus, ONERA, and DLR Results from the 2nd AIAA Drag Prediction 38 Brodersen, O., “Numerische Analyse der aerodynamischen Triebw-
Workshop,” AIAA Paper 2004-0391, Jan. 2004. erksinstallationseffekte an Transportflugzeugen,” DLR, German Aerospace
12 Langtry, R., Kuntz, M., and Menter, F., “Drag Prediction of Engine- Research Center, DLR Technical Rept. 2003-10, Cologne, Germany,
Airframe Interference Effects with CFX 5,” AIAA Paper 2004-0392, May 2003.
Jan. 2004. 39 Vassberg, J., DeHaan, M., and Sclafani, T., “Grid Generation Require-
13 Sclafani, A., DeHaan, M., and Vassberg, J., “OVERFLOW Drag Pre- ments for Accurate Drag Predictions Based on OVERFLOW Calculations,”
diction for the DLR-F6 Transport Configuration: A DPW-II Case Study,” AIAA Paper 2003-4124, June 2003.
AIAA Paper 2004-0393, Jan. 2004. 40 Roache, P. J., “Verification and Validation in Computational Science
14 Rumsey, C., Rivers, S., and Morrision, J., “Study of CFD Variation and Engineering,” Hermosa, Albuquerque, NM, 1998.
on Transport Configurations from the Second Drag-Prediction Workshop,” 41 Tinoco, E. N., “An Assessment of CFD Prediction of Drag and Other
AIAA Paper 2004-0394, Jan. 2004. Longitudinal Characteristics,” AIAA Paper 2001-1002, June 2002.