10+2 common misconceptions about aerodynamics

This list is a summary of misinterpretations that I have noticed during the last years as an aeronautical engineer, which from my perspective (and probably in the opinion of the editors and reviewers who have understood and published my research) do not allow to know the fundamental physics of real aerodynamics (and fluid dynamics). As usual in this blog, I try to be concise by developing my explanation for each of them in a single paragraph. No more blah, blah, blah and let's get to the point.

Fig. 1 Continuity exists everywhere. Sky and Water (M.C. Escher, 1938).

1. "Ideal flows perfectly attach to the surfaces".

Although it is not yet possible to confirm or deny this from an experimental point of view (the superfluid fountain is not a flow past an object experiment), the Potential Flow Theory (PFT) assumes that an ideal (i.e. incompressible, irrotational and inviscid; i-i-i) flow will turn around, even with sharp leading edges, to maintain a perfectly attached flow pattern to the body contour without any separation. However, such a behavior resembles a viscous fluid pattern (i.e., the opposite case) within the low Reynolds number (Re) condition, while the PFT is theoretically established in the highest limit, i.e., infinite Re, where all flow variables are assumed instantaneously transported. 

Fig. 2 Clearly, these solutions do not represent potential (i-i-i) flow simulations, since the particles do not move with 'infinite' velocity (according to the characteristic length), but all the opposite; linearized NSE do not represent ideal flow solutions!

Also see: A four-question quiz: on fundamentals of fluid dynamics (librepenzzzador.blogspot.com)

2. "Inviscid flows are vorticity-free".

By definition, fluid viscosity is a property that naturally diffuses (gradually reduces with time) vorticity; from this point of view, an inviscid flow past a body must be riddled with vortices, as long as they have been previously generated by some physical mechanism (see inviscid vorticity generation theories [Morton, 1984; Terrington et al., 2022]). As several authors (G.K. Batchelor, J. Katz, P.G. Saffman, etc.) point out, in an ideal flow past a body, all vorticity is confined to an infinitely thin boundary layer (BL), since the flow velocity is assumed to be infinite (the thickness of the BL decreases as the flow velocity increases), so such a case is not strictly vorticity-free, hence rotational at the surface!

Also see: On detached potential flow and the d'Alembert's paradox (September-2022) (librepenzzzador.blogspot.com)

3. "Potential flow models a free-slip boundary condition at surface".

This means that there is no physical contact (no shear) between the flow and the body surface, so there should be no reaction force between them. Furthermore, if this were the case, no vorticity could be generated at the body-flow interface, which would become just another flow-flow interface like in the remaining flow layers around the object; this assumption directly contradicts the interpretation of the existence of an infinitely thin BL at the body surface.

Also see: A fundamental question in fluid dynamics: vorticity generation (librepenzzzador.blogspot.com)

4. "Vorticity is only detached from external edges (e.g., wing tips and trailing edge)".

The Kutta condition (e.g., at the trailing edge of an airfoil) forces the flow (circulation-vorticity) to remain attached to the surface, confined within an infinitely thin BL, which is clearly a viscous fluid past a body behaving. Thus, the perpendicular projection (lift) of the resultant force obtained by the Circulation Theory of Lift (by Kutta and Zhukovskiy; KJ) for an "ideal" flow past a body includes both inviscid and viscous contributions, which is why such results agree well with experimental and viscous numerical data (for airfoils, wings, and aircraft operating in the high Re range) without any BL coupling (it could help to improve the real BL thickness due to a lower than infinite Re).

Also see: Can vorticity only be generated along external separation lines? (May-2022) (librepenzzzador.blogspot.com)

5. "Viscosity leads to fluid separation".

This idea has deep roots even among Navier-Stokes (NS) fluid dynamists, who seem to be inspired by the belief that inviscid flows exist perfectly attached to surfaces (without any physical contact due to a supposed free-slip condition), so that the appearance of viscosity leads to fluid separation, although this is counterintuitive in itself. By definition, viscosity is a property that allows a fluid to adhere to surfaces (due to friction between its layers), but not the other way around. Thus, by inverse logic, a hypothetical inviscid flow should allow complete separation (due to the absence of friction between its layers), allowing vorticity to be detached from the entire surface, thus avoiding any BL formation.

Also see: More patents, less papers (librepenzzzador.blogspot.com)

6. "The leading edge does not contribute to lift"

Experimental research has shown that the leading edge (LE) vortex (LEV), which remains attached to the aerodynamic body under certain operating conditions (e.g. pre-stall), also contributes to the aerodynamic force, hence lift. This fact can be justified through the typical pressure distribution around an airfoil at relatively low AoA, which shows that the maximum pressure coefficient (suction peak) decays abruptly around the frontal stagnation point due to the LE curvature. Even in a flat plate (with a sharp LE), lift tends to be concentrated towards the LE at relatively small AoA, while at higher AoA the pressure is better distributed along and across its suction side (see Fig. 3). Some semi-empirical (e.g., Polhamus suction analogy) and analytical models have been implemented to try to account for the contribution of the LEV in simplified geometries. But, what are the limitations of obtaining it from a purely numerical approach? I have not seen any [1].

Fig. 3 Pressure distribution on the suction side for a square flat plate (with beveled edges) at low and high AoAs (Winter, 1936).

Also see: How does the stall mechanism occurs? (librepenzzzador.blogspot.com)

7. "Lift is a force".

Neither lift nor drag are forces per se, but the perpendicular and parallel projections of a resulting aerodynamic force vector. This fact is relevant when some authors try to explain how lift (or drag) is generated (KJ for lift, and Prandtl for drag), but in an independent way, as if each were a force in itself. Any theory constructed under such a premise should be incomplete. A complete theory (first, for incompressible fluid flow) must explain how the (total) resulting force is generated, but not only for an attached flow assumption, but also for a detached one (even for both inviscid and viscous cases).

Also see: What is lift "force"? NASA is wrong! (librepenzzzador.blogspot.com)

8. "Lift is due to pressure difference between upper and lower part".

First we have to define pressure: a perpendicular FORCE per unit area. So, how can the existence of the lift "force" be justified by a simple pressure difference (after all, pressures are only distributed FORCES!)? In other words, "a distributed force try to justify another force"; this is simplistic in a more formal context and does not help to answer the question of the origin of such a force. The question should be: Where do such 'distributed forces' come from? and the real answer should be to look for a justification for the existence of such distributed forces (pressures). Velocity field is the answer; velocity causes pressures and shear (tangential) stresses, which also contribute to lift under some operating conditions! (i.e., an airfoil at low AoA range). So, in the strictest sense, not only "(perpendicular) PRESSURE difference" causes lift. The velocity (and vorticity) field causes it all!

Also see: Does vorticity "induces" velocity? (librepenzzzador.blogspot.com)

9. "Fluid separation is due to the 'adverse pressure gradient'".

The Bernoulli equation is a simplified version (for incompressible, inviscid flow, among other assumptions) of the NS equations that allows to know the direct relationship between the flow velocity and its pressure. In such an equation, the part corresponding to the calculation of the dynamic pressure shows that the independent term is the velocity (u), while the dependent one is the dynamic pressure (q), i.e. low velocity causes high dynamic pressure and high velocity causes low dynamic pressure. Thus, it is inaccurate to say that pressure gradients (adverse or favorable) cause flow separation or attachment; the cause is the developed velocity field (besides, this explanation is more intuitive: flow particles move independently according to an instantaneous velocity field that causes a variation in the surrounding pressure). This is clear in the velocity-vorticity formulation of the NS equations, where the pressure field is decoupled from the flow solution (it can be reconstructed a posteriori).

Also see: No more fallacies. Why do airplanes fly? (librepenzzzador.blogspot.com)

10. "The boundary layer is an entity appart, thus it must be treated in a special way".

The concept of a "boundary layer" (laminar, transitional and turbulent) is useful to explain how shear stresses act near a solid surface in a viscous fluid, and then to model such a region. However, since its physical nature (fluid particles) is exactly the same as the rest of the fluid (viscous-inviscid continuity condition is gradual but not abrupt), its existence from a fluid particle perspective is reduced to a purely conceptual idea. This is clearly demonstrated by a Lagrangian solution of the NS equations, where the thickness and velocity profile near a solid surface (flat plate) is numerically approximated without any additional assumption or modeling of the BL (see chapter Boundary layers by discrete vortex modelling in [2]).

Fig. 4 Blasius vs. vortex cloud theory (numerical). Source: [2]

Fig. 5 Horizontal and vertical components of velocity within BL (numerical vs analytical). Source: [3]

Also see: Can a brick 'fly' (glide)? (librepenzzzador.blogspot.com)

10+1. "Turbulence models are the Holy Grail of CFD".

At this point, the Reynolds Averaged Navier-Stokes (RANS) method is still the most practical way to approximate the simplest fluid simulations, since it introduces a set of semi-empirical models to try to close such a mathematical system of equations. Of course, it is far from being the optimal way to solve fluid dynamics numerically, since it depends on experimental coefficients chosen according to the physical conditions to be simulated. For this reason, simpler (in terms of assumptions) mesh-based methods (e.g., LES and DNS) have been developed to avoid such artificial dependence to some extent, but they require much more computational resources. On the other hand, vorticity-based methods theoretically do not require turbulence models!

Fig. 6 List of various turbulence models available. Source:[PDF] Guidelines for choosing an appropriate turbulence model when simulating blood flows in cardiovascular devices | Semantic Scholar

Also see: What is the 'hybrid vortex tube-vorton' method? (librepenzzzador.blogspot.com)

10+2. "DNS is the panacea of fluid (and flow) dynamics".

Aside from its current impracticality due to the extremely high computational resources required, mesh-based direct numerical simulation (DNS) is not yet able to answer at least two fundamental aspects of fluid-flow dynamics: 1) Can vorticity be generated in an inviscid flow? 2) Can an external force be generated in an inviscid medium? DNS is unable to provide such answers, since it solves the NS equations (by including the viscous term), but not the Euler equations (inviscid case). That is, no numerical solutions exist for a fluid viscosity value close to zero (or Re tending to infinity) due to its current inherent spatial and temporal limitation in the upper Re limit.

Also see: Aerodynamics is just a philosophy (librepenzzzador.blogspot.com)

[1] The Full Multi-wake Vortex Lattice Method: a detached flow model based on Potential Flow Theory | Advances in Aerodynamics | Full Text (springeropen.com)

[2] Lewis, R.I. (1991). Vortex Element Methods for Fluid Dynamic Analysis of Engineering Systems. Cambridge Engine Technology Series.

[3] A. Kolganova, I. Marchevsky, and E. Ryatina (2023), Hybrid Barnes–Hut/Multipole algorithm application to vortex particles velocities calculation and integral equation solution, J. Phys.: Conf. Ser. 2543, Singapore.

"If we start making assumptions and simplifications too early, we may miss the big picture that is essential to solving the whole problem".