You are on page 1of 7

Benchmarking the M2DO FEA Doublet Lattice

Method
August 16, 2016

Summary

Over the 3 weeks, I have worked on implementing the doublet lattice method
(DLM) as a physics module in the M2DO FEA c++ code. There are two components to the DLM: the steady (horshoe vortex) part and the unsteady (oscillatory) part. Both parts have now been implemented and used to produce
results for several benchmark problems; I consider them to be within acceptable
accuracy (see discussion below). I will now continue to produce documentation
for how the DLM physics works and move on to combine the DLM with the
structural mechanics module (already implemented) to produce and solve the
flutter equation.

Theoretical References

I utilized two main sources in order to program the DLM physics: A 1971
paper by Giesling et al [1] and a text book by Katz [2]. Both ASTROS [3] and
NASTRAN [4] reference the Giesling [1] paper; the Katz [2] textbook was used
as it contains more elementary explanations of the origins of the DLM theory.
Results within these texts and also two more papers, by Rodden et al [5]
and Albano et al [6] repectively, were used as benchmarks against which the
M2DO FEA result could be compared (see below).

Benchmark Results for Steady DLM

The textbook by Katz [2] contained a few useful results graphs, which combined
the authors implementation with some experimental results, and also some
results by Rodden. Unfortunately, no actual numerical values were given, so
below in Fig. 1 I have overplotted my results (in blue) on top of the graph
images given in [2] (in black).
Fig. 1a depicts the effect of sweep angle on lift coefficient slope (CL, ).
Fig. 1b shows the distribution of loading along the length of the wing. Fig. 1c

depicts the effect of dihedral angle and also ground interaction on the lift coefficient slope.
Clearly, the trend calculated using the M2DO FEA code is visually very similar
to that reported in the literature. There are a few places where the results
are not in strict agreement, such as for very large negative dihedral angles in
Fig. 1c. To be honest, I am not sure why this is the case. It could be a
small implementation error on my part; If so, once I have finished writing the
documentation, myself or someone else in the group will likely spot the error. It
could also just be a slight difference in the way angle of attack is implemented,
for instance, which would mean that neither is more correct.
In any case, as we can see, the overall trends are definitely captured by the
code in its current form; in light of this I feel happy to proceed to the unsteady
calculations.

Results Visualization

I have also built into the code a way to visualize the flow around the analyzed
wing. Since the DLM only solves for pressure coefficient (thus vorticity) on the
surface, I used a secondary (volume) mesh to extend the results from the surface
into the space near the wing. Once we have a vector (flow) field in this space,
Paraview can be used to plot streamlines, revealing the flow characteristics.
An example of this functionality is shown in Fig. 2, where I have analzed a
swept, twisted wing at 10 degrees angle of attack.

Benchmark Results for Unsteady DLM

The papers by Rodden eta al [5] and Albano et al [6] contained results for a
range of wing configurations, which could serve as a benchmark set for unsteady
DLM calculations. As above, no actual numerical values were given, and so in
figures 3 and 4 I have overplotted my results (in blue) over the respective authors
(in black). As in the steady case above, the results are in general agreement.

Next Steps

The result of the DLM is that pressure loading on a wing can be serived, given
the wing shape and deformation. Via a surface spline, the loading can be transferred to a mechanical model, thus forming the flutter equations.
The next step is thus to generate a flutter equation, solve it, and compare
the results to those from ASTROS.

(a) Sweep angle vs CL,

(b) Sweep angle vs spanwise loading

(c) Dihedral angle and ground effect

Figure 1: Comparison of results given in [2] with those obtained with M2DO FEA.
3

Figure 2: Stream lines near a swept, twisted lifting surface. Color indicates
strength of local vorticity.

(a) <{K}, AR = 2.0

(b) <{K}, AR = 5.0

(c) <{CL, }, AR = 20.0

(d) ={CL, }, AR = 20.0

Figure 3: Comparison of results given in [5] with those obtained with M2DO FEA.

(a) Lift distribution near root.

(b) Lift distribution near tip.

Figure 4: Lift distribution on rectangular AR = 3.0 wing in bending mode;


comparison of results given in [6] with those obtained with M2DO FEA.

References
[1] JP Giesing, TP Kalman, and WP Rodden. Subsonic unsteady aerodynamics
for general configurations. Air Force Flight Dynamics Laboratory Report
AFFDL-TR-71-5, 1971.
[2] Joseph Katz and Allen Plotkin. Low-speed aerodynamics, volume 13. Cambridge University Press, 2001.
[3] DJ Neill, DL Herendeen, and VB Venkayya. Astros enhancements: Volume
iii-astros theoretical manual. Flight Dynamics Directorate, Wright Laboratory, Wright Patterson Air Force Base, WL-TR-95-3006, 1995.
[4] Richard H MacNeal. The NASTRAN theoretical manual, volume 221. Scientific and Technical Information Office, National Aeronautics and Space
Administration, 1970.
[5] William P Rodden, Paul F Taylor, and Samuel C McIntosh. Further refinement of the subsonic doublet-lattice method. Journal of Aircraft, 35(5):720
727, 1998.
[6] Edward Albano and William P Rodden. A doublet-lattice method for calculating lift distributions on oscillating surfaces in subsonic flows. AIAA
journal, 7(2):279285, 1969.

You might also like