Fluids & Structural Mechanics > CM Projects > Drag Reduction and Flow Control (DRFC)

Images (Below)

The US Navy has pursued a variety of technologies to reduce drag and/or improve the hydrodynamic performance of hulls and control surfaces. ARL’s Computational Mechanics Division and Fluid Dynamics Division have studied additive drag reduction technologies computationally and experimentally for decades.

Between 2002 and 2006, DARPA’s Friction Drag Reduction (FDR) program was a major sponsor of ARL’s development of CFD modeling for microbubble drag reduction (MBDR). In this research, an Eulerian 2-fluid approach was taken to enable analysis of very high (i.e., Navy relevant) Reynolds number flows with MBDR. In this approach, the various constituents are treated as inter-penetrating, but are not assumed to be in dynamic equilibrium. Specifically, the bubbles of various sizes travel at different velocities than the surrounding water. Accordingly gas-liquid interface forces arises that must be modeled. Models for drag, virtual mass, lift and turbulence dispersion are of particular importance in these flows. These models are particularly challenging to formulate and validate due to the high gas volume fractions involved (local peak volume fractions can exceed 0.5, i.e., disperse flow assumptions are invalid). A range of bubble sizes exist in these flows arising from the equilibrium between coalescence and breakup that arises; these two mass transfer mechanisms are modeled as well.

An in-house developed multi-phase CFD code, NPHASE-PSU was instrumented with the various physical models required to attack MBDR flows. The aforementioned models were implemented and evolved as a range of validation data became available. In particular DNS simulations were performed by colleagues (and DARPA FDR program awardees) at Brown University and Worchester Polytechnic Institute. Figure 1 shows an example comparison between NPHASE-PSU and DNS for the simulation of a Re* ≈ 377 channel flow with center peaked inlet bubble distribution. This low Reynolds number collaboration/validation was critical in our team’s development of a new turbulence dispersion model that accounts for homogeneous turbulence as well as bubble-collision-induced dispersion effects. At higher Reynolds numbers, first-of-its-kind experimental measurements of bubble volume fraction profiles in a flat plate turbulent boundary layer were obtained in ARL’s 12” water tunnel by collaborators in Fluid Dynamics. Figure 2a and Figure 2b shows a representative comparison between NPHASE-PSU predictions and this moderate Reynolds number (Rex symbol 10x106) data. Very high Reynolds number measurements were obtained by colleagues (and DARPA FDR program awardees) at the University of Michigan. The 10m long HIPLATE was instrumented with MBDR injection (see Figure 3) and installed in the Naval Surface Warfare Center’s Large Cavitation Channel in Memphis, TN. Skin friction, near-wall gas volume fraction, near-wall bubble sizes and near-wall bubble velocities were obtained at Reynolds numbers approaching 3x108. Figures 4a and 4b show representative comparisons between NPHASE-PSU predictions and this high Reynolds number (Rexsymbol 10x106) data. Further details of this program can be found in (Kunz et al. 2007).

Flow control can be defined as active or passive mechanisms employed to enhance the performance of aerodynamic/hydrodynamic surfaces. The many approaches that have been studied aim to reduce drag and/or increase lift of aerodynamic surfaces. Under sponsorship from the US Office of Naval Research (ONR) ARL’s Computational Mechanics Division has studied Coanda Effect Circulation Control (CC). In these approaches, steady or unsteady blowing is introduced near the foil trailing edge. With a curved trailing edge surface, the Kutta condition does not apply, and the rear stagnation point is free to move. The result is a net change in the circulation, with the flow direction and separation location can be altered by changing the rate of blowing. Potential benefits of CC for undersea systems include improved control authority at lower speeds and high angle of attack, as arise in littoral operations and evasive maneuvers. Figure 5 illustrates the concept.

In order to resolve the physics of the multiple high speed shear layers in these systems (injection velocities approaching sonic in air applications), high quality computational meshes much be constructed, especially in the vicinity of injection. Our work focused on applying structured overset grid generation methods (see Computational Mechanic's Overset Grid Methods) and unstructured methods as illustrated in Figure 6. Turbulence modeling is also important in these flows. Figure 7 shows two views of a Detached Eddy Simulation (DES) of a CC airfoil illustrating the complex vortical field that arises in the wake of the relatively blunt airfoil analyzed. Figure 8 shows a comparison of predicted and measured near-surface velocity profiles along the trailing edge of a CC airfoil.

Images
image
View Image (25kb)

Figure 1:

High volume fraction comparison between NPHASE-PSU and DNS for the simulation of a Re* ≈ 377 channel flow with center peaked inlet bubble distribution.

image
View Image (36kb)

Figure 2a:

Comparison between NPHASE-PSU and 12-inch tunnel measurements. Void fraction profiles at two axial locations for tunnel velocity of U=13.7 m/s, gas injection rate of Q=2.1 l/s.

image
View Image (29kb)

Figure 2b:

Comparison between NPHASE-PSU and 12-inch tunnel measurements. Drag reduction at a downstream axial location for three tunnel velocities and three gas injection rates.

image
View Image (109kb)

Figure 3:

Illustration of HIPLATE configuration.

image
View Image (42kb)

Figure 4a:

Comparison between NPHASE-PSU and HIPLATE measurements. Predicted and measured drag reduction vs. x for Uref=12 m/s at three gas injection rates.

image
View Image (27kb)

Figure 4b:

Comparison between NPHASE-PSU and HIPLATE measurements. Predicted and measured bubble velocity vs. volumetric fraction of gas flow rate at upstream and downstream measurement stations for Uref=12 and 18 m/s at three injection rates.

image
View Image (106kb)

Figure 5:

Trailing edge detail for a 2D Coanda Effect Circulation Control scheme.

image
View Image (166kb)

Figure 6:

Structured overset and unstructured grid topologies employed for Coanda Effect Circulation Control simulations.

image
View Image (266kb)

Figure 7:

Side and top views of a Detached Eddy Simulation (DES) of a CC airfoil illustrating the complex vortical field that arises in the wake of the relatively blunt airfoil analyzed. Shown are isosurfaces of intrinsic-swirl colored by normalized helicity.

image
View Image (142kb)

Figure 8:

Comparison of predicted and measured near-surface velocity profiles along the trailing edge of a CC airfoil. Measurement locations indicated in top figure.