Aerodynamic potential-flow code

In fluid dynamics, aerodynamic potential flow codes or panel codes are used to determine the fluid velocity, and subsequently the pressure distribution, on an object. This may be a simple two-dimensional object, such as a circle or wing, or it may be a three-dimensional vehicle.

A series of singularities as sources, sinks, vortex points and doublets are used to model the panels and wakes. These codes may be valid at subsonic and supersonic speeds.

History

Early panel codes were developed in the late 1960s to early 1970s. Advanced panel codes, such as Panair (developed by Boeing), were first introduced in the late 1970s, and gained popularity as computing speed increased. Over time, panel codes were replaced with higher order panel methods and subsequently CFD (Computational Fluid Dynamics). However, panel codes are still used for preliminary aerodynamic analysis as the time required for an analysis run is significantly less due to a decreased number of elements.

Assumptions

These are the various assumptions that go into developing potential flow panel methods:

• Inviscid
• Incompressible $\nabla \cdot V=0$
• Irrotational $\nabla \times V=0$
• Steady ${\frac {\partial }{\partial t}}=0$

However, the incompressible flow assumption may be removed from the potential flow derivation leaving:

• Potential flow (inviscid, irrotational, steady) $\nabla ^{2}\phi =0$

Derivation of panel method solution to potential flow problem

• From Small Disturbances

$(1-M_{\infty }^{2})\phi _{xx}+\phi _{yy}+\phi _{zz}=0$ (subsonic)

• From Divergence Theorem

$\iiint \limits _{V}\left(\nabla \cdot \mathbf {F} \right)dV=\iint \limits _{S}\mathbf {F} \cdot \mathbf {n} \,dS$

• Let Velocity U be a twice continuously differentiable function in a region of volume V in space. This function is the stream function $\phi$.
• Let P be a point in the volume V
• Let S be the surface boundary of the volume V.
• Let Q be a point on the surface S, and $R=|P-Q|$.

As Q goes from inside V to the surface of V,

• Therefore:

$U_{p}=-{\frac {1}{4\pi }}\iiint \limits _{V}\left({\frac {\nabla ^{2}\cdot \mathbf {U} }{R}}\right)dV_{Q}$

$-{\frac {1}{4\pi }}\iint \limits _{S}\left({\frac {\mathbf {n} \cdot \nabla \mathbf {U} }{R}}\right)dS_{Q}$

$+{\frac {1}{4\pi }}\iint \limits _{S}\left(\mathbf {U} \mathbf {n} \cdot \nabla {\frac {1}{R}}\right)dS_{Q}$

For :$\nabla ^{2}\phi =0$, where the surface normal points inwards.

$\phi _{p}=-{\frac {1}{4\pi }}\iint \limits _{S}\left(\mathbf {n} {\frac {\nabla \phi _{U}-\nabla \phi _{L}}{R}}-\mathbf {n} \left(\phi _{U}-\phi _{L}\right)\nabla {\frac {1}{R}}\right)dS_{Q}$

This equation can be broken down into both a source term and a doublet term.

The Source Strength at an arbitrary point Q is:

$\sigma =\nabla \mathbf {n} (\nabla \phi _{U}-\nabla \phi _{L})$

The Doublet Strength at an arbitrary point Q is:

$\mu =\phi _{U}-\phi _{L}$

The simplified potential flow equation is:

$\phi _{p}=-{\frac {1}{4\pi }}\iint \limits _{S}\left({\frac {\sigma }{R}}-\mu \cdot \mathbf {n} \cdot \nabla {\frac {1}{R}}\right)dS$

With this equation, along with applicable boundary conditions, the potential flow problem may be solved.

Required boundary conditions

The velocity potential on the internal surface and all points inside V (or on the lower surface S) is 0.

$\phi _{L}=0$

The Doublet Strength is:

$\mu =\phi _{U}-\phi _{L}$

$\mu =\phi _{U}$

The velocity potential on the outer surface is normal to the surface and is equal to the freestream velocity.

$\phi _{U}=-V_{\infty }\cdot \mathbf {n}$

These basic equations are satisfied when the geometry is a 'watertight' geometry. If it is watertight, it is a well-posed problem. If it is not, it is an ill-posed problem.

Discretization of potential flow equation

The potential flow equation with well-posed boundary conditions applied is:

$\mu _{P}={\frac {1}{4\pi }}\iint \limits _{S}\left({\frac {V_{\infty }\cdot \mathbf {n} }{R}}\right)dS_{U}+{\frac {1}{4\pi }}\iint \limits _{S}\left(\mu \cdot \mathbf {n} \cdot \nabla {\frac {1}{R}}\right)dS$

• Note that the $dS_{U}$ integration term is evaluated only on the upper surface, while th $dS$ integral term is evaluated on the upper and lower surfaces.

The continuous surface S may now be discretized into discrete panels. These panels will approximate the shape of the actual surface. This value of the various source and doublet terms may be evaluated at a convenient point (such as the centroid of the panel). Some assumed distribution of the source and doublet strengths (typically constant or linear) are used at points other than the centroid. A single source term s of unknown strength $\lambda$ and a single doublet term m of unknown strength $\lambda$ are defined at a given point.

$\sigma _{Q}=\sum _{i=1}^{n}\lambda _{i}s_{i}(Q)=0$

$\mu _{Q}=\sum _{i=1}^{n}\lambda _{i}m_{i}(Q)$

where:

$s_{i}=ln(r)$

$m_{i}=$

These terms can be used to create a system of linear equations which can be solved for all the unknown values of $\lambda$.

Methods for discretizing panels

• constant strength - simple, large number of panels required
• linear varying strength - reasonable answer, little difficulty in creating well-posed problems
• quadratic varying strength - accurate, more difficult to create a well-posed problem

Some techniques are commonly used to model surfaces.

• Body Thickness by line sources
• Body Lift by line doublets
• Wing Thickness by constant source panels
• Wing Lift by constant pressure panels
• Wing-Body Interface by constant pressure panels

Methods of determining pressure

Once the Velocity at every point is determined, the pressure can be determined by using one of the following formulas. All various Pressure coefficient methods produce results that are similar and are commonly used to identify regions where the results are invalid.

Pressure Coefficient is defined as:

$C_{p}={\frac {p-p_{\infty }}{q_{\infty }}}={\frac {p-p_{\infty }}{{\frac {1}{2}}\rho _{\infty }V_{\infty }^{2}}}={\frac {p-p_{\infty }}{{\frac {\gamma }{2}}p_{\infty }M_{\infty }^{2}}}$

The Isentropic Pressure Coefficient is:

$C_{p}={\frac {2}{\gamma M_{\infty }^{2}}}\left(\left(1+{\frac {\gamma -1}{2}}M_{\infty }^{2}\left[{\frac {1-|{\vec {V}}|^{2}}{|{\vec {V_{\infty }}}|^{2}}}\right]\right)^{\frac {\gamma }{\gamma -1}}-1\right)$

The Incompressible Pressure Coefficient is:

$C_{p}=1-{\frac {|{\vec {V}}|^{2}}{|{\vec {V_{\infty }}}|^{2}}}$

The Second Order Pressure Coefficient is:

$C_{p}=1-|{\vec {V}}|^{2}+M_{\infty }^{2}u^{2}$

The Slender Body Theory Pressure Coefficient is:

$C_{p}=-(2u+v^{2}+w^{2})$

The Linear Theory Pressure Coefficient is:

$C_{p}=-2u$

The Reduced Second Order Pressure Coefficient is:

$C_{p}=1-|{\vec {V}}|^{2}$

What panel methods cannot do

• Panel methods are inviscid solutions. You will not capture viscous effects except via user "modeling" by changing the geometry.
• Solutions are invalid as soon as the flow changes locally from subsonic to supersonic (i.e. the critical Mach number has been exceeded) or vice versa.

Potential flow software

See also: Computational fluid dynamics § Background and history

Name	License	Lan	Operating system			Geometry import	Meshing			Body Representation	Wake model	Developer
			Linux	OS X	Microsoft Windows		Structured	Unstructured	Hybrid
Aeolus ASP	Proprietary	Java / Fortran	Yes		Yes		Yes			Quadrilaterals		Aeolus Aero Sketch Pad
CMARC	Proprietary	C	Yes		Yes							Homepage, AeroLogic, based on PMARC-12
DesignFOIL	Proprietary		Wine		Yes							www.dreesecode.com, DreeseCode Software LLC
FlightStream	Proprietary	Fortran / C++			Yes	CAD, Discrete		Yes	Yes	Solids		Research in Flight Company
HESS	Proprietary											Douglas Aircraft Company
LinAir	Proprietary		Yes									Desktop Aeronautics
MACAERO	Proprietary											McDonnell Aircraft
NEWPAN	Proprietary	C++	Yes				Yes	Yes				Flow Solutions Ltd.
Tucan	GPLv3	VB.NET/C#.NET	Yes (Console)		Yes	STL	Yes			Quadrilaterals & triangles	Free	G. Hazebrouck & contributors
QBlade	GPLv2	C/C++	Yes									TU Berlin
Quadpan	Proprietary											Lockheed
PanAir a502	Public domain software	Fortran	Yes	Yes	Yes							Homepage, Boeing?
PANUKL	freeware	C++/Fortran	Yes		Yes	NX - partially	Yes			Quadrilaterals		Warsaw University of Technology, PANUKL exports data to SDSA and to Calculix
PMARC	Free On Request	Fortran 77	UNIX	Yes	Yes							NASA, descendant of VSAERO
VSAero	Proprietary		UNIX									Homepage
Vortexje	GPLv2	C++	Yes									Baayen & Heinz GmbH
XFOIL	GPLv2	Fortran	Yes	Yes	Yes							web.mit.edu/drela/Public/web/xfoil/
XFLR5	GPLv2	C/C++	Yes									www.xflr5.com
VSPAERO Packaged with OpenVSP	NOSA	C++	Yes	Yes	Yes			Yes		Polygons, typically quad & tri dominated	Free & rigid	openvsp.org
MachLine	MIT License	Fortran	Yes	untested	untested	STL, VTK, TRI		Yes		Solid bodies using surface tris	Rigid	Utah State University AeroLab aerolab.usu.edu github.com/usuaero/MachLine

See also

• Stream function
• Conformal mapping
• Velocity potential
• Divergence theorem
• Joukowsky transform
• Potential flow
• Circulation
• Biot–Savart law

References

• Public Domain Aerodynamic Software, A Panair Distribution Source, Ralph Carmichael
• Panair Volume I, Theory Manual, Version 3.0, Michael Epton, Alfred Magnus, 1990 Boeing
• Panair Volume II, Theory Manual, Version 3.0, Michael Epton, Alfred Magnus, 1990 Boeing
• Panair Volume III, Case Manual, Version 1.0, Michael Epton, Kenneth Sidewell, Alfred Magnus, 1981 Boeing
• Panair Volume IV, Maintenance Document, Version 3.0, Michael Epton, Kenneth Sidewell, Alfred Magnus, 1991 Boeing
• Recent Experience in Using Finite Element Methods For The Solution Of Problems In Aerodynamic Interference, Ralph Carmichael, 1971 NASA Ames Research Center
• [1]