12-11-2012, 03:30 PM
Direct Numerical Simulation of Swept-Wing Laminar Flow Control Using Pinpoint Suction
Direct Numerical Simulation.pdf (Size: 3.23 MB / Downloads: 139)
Introduction
Improving the fuel efficiency of aircrafts has become an important task within
the last decades. Not only do airlines benefit from saving fuel by decreasing
their direct operation costs regarding the oil price development but also the
environmental aspect has gained growing interest and it will only be a matter
of time until environmental laws limiting exhaust gases will be approved.
Current commercials for newly designed aircrafts are inconceivable without
sentences like “The 787 Dreamliner is using 20 percent less fuel than any other
airplane of its size” (Boeing homepage).
Until today actually applied optimisations for new airplanes are limited
to enhanced shaping, higher surface quality and engine improvement, but
little potential is thought to be left in these research fields. New concepts
have therefore to be envisaged. Laminar flow control (LFC) provides a total
drag reduction potential of up to 16 % by e.g. realizing 40% laminar flow on
wings and control surfaces of a current airliner.
Compressible Code NS3D
The DNS code NS3D (see Babucke, Linn, Kloker, and Rist [1]) is used for
the compressible test simulations. This code is based on the complete 3-d
unsteady compressible Navier-Stokes equations and a calorically perfect gas.
The equation set is also solved in a rectangular integration domain on the flat
plate. In streamwise (x-) and wall-normal (y-) direction, the discretization is
realized by splitted compact finite differences of 6th order. In the spanwise
(z-) direction, the flow is assumed to be periodic, thus a Fourier spectral
representation is employed to compute the z-derivatives. In contrast to the incompressible code N3D, NS3D largely computes in physical space. After
transformation to Fourier space and simple computation of the z-derivatives,
the back transformation is done with de-aliasing using the 2/3-rule. For time
integration the classical 4th-order Runge-Kutta method is employed as in the
incompressible case. A detailed description of the discretization algorithm and
boundary conditions is given in Babucke & al. [1] and Linn & Kloker [10].
Results
For the incompressible code N3D a strong dependence of the necessary wallnormal
resolution at the wall, Δywall, and the maximum suction amplitude
prescribed at the hole was found. In order to confirm the results with very
high suction amplitudes, a grid-study test case was simulated with the incompressible
and compressible codes N3D and NS3D, respectively, with varying
Δywall. A single suction hole with 30% maximum suction velocity in a twodimensional
Blasius boundary layer flow (U∞=15.0m
s ) was chosen. All remaining
flow parameters were matched to Meitz’s case of Goldstein [11]. Since for
the compressible code the allowable time step is roughly proportional to the
Mach number, U∞ of the compressible boundary layer was chosen to be 4.6
times higher in order to achieve Ma=0.2 and thus a reasonable time step. In
order to allow for comparison, Reδ1 (xhole), dhole/δ1 and dhole/sz have been
kept constant for both codes, where dhole is the hole diameter, δ1 the local
displacement thickness, and sz = λz,0 is the spanwise spacing of the holes and
equal to the fundamental spanwise wavelength. The parameters for both simulations
can be found in Table 1.