Rankine Oval Dynamics

• Rankine oval is a closed stagnation streamline generated by superimposing a uniform stream with a source–sink pair, forming an exact potential-flow body.
• The geometry is controlled by the nondimensional parameter Ua/m, which governs the shape transition from circular to elongated ovals and influences drag characteristics.
• In viscous flow, the Rankine oval behaves as a bluff body, experiencing boundary-layer separation, wake instability, and periodic vortex shedding.

The Rankine oval is the closed stagnation streamline obtained by superimposing a uniform stream of speed $U_\infty$ with a source–sink pair of equal and opposite strength $m$, separated by a distance $2a$. In classical potential flow it is an exact irrotational construction and is therefore “streamlined” in the inviscid sense; in viscous flow, however, the same geometry behaves as a bluff body, exhibiting boundary-layer separation, recirculation, wake instability, and eventually periodic vortex shedding. A systematic direct numerical simulation study of incompressible flow past Rankine ovals over $Re=10$ to $200$ and $Ua/m=0$ to $1$ establishes how the shape parameter $Ua/m$ controls the transition from steady to periodic wakes, the scaling of drag and shedding frequency, and the extent to which potential-flow information remains useful in viscous drag prediction (Xu et al., 8 Sep 2025).

1. Classical definition and exact construction

In the standard construction, a source of strength $+m$ is placed at $(x,y)=(-a,0)$, a sink of strength $m$0 at $m$1, and a uniform stream of velocity $m$2 is imposed in the $m$3-direction. The body boundary is the streamline $m$4, with streamfunction

$m$5

In Cartesian form, the same boundary is written as

$m$6

The paper also gives an implicit dimensionless relation. The typesetting in the manuscript is corrupted, but the intended form is reported as

$m$7

The governing geometric control parameter is the nondimensional quantity

$m$8

abbreviated as $m$9. Physically, it is the inverse of a nondimensional source strength. Small $2a$0 corresponds to stronger source–sink influence relative to the freestream and produces fuller, more circular bodies; large $2a$1 corresponds to stronger freestream influence and produces longer, more slender ovals (Xu et al., 8 Sep 2025).

The Rankine oval is therefore not merely a descriptive shape class but an exact streamline body generated by a specific potential-flow superposition. This exactness is central to its historical role in inviscid theory and to its later reinterpretation as a viscous bluff-body problem.

2. Shape parameterization, normalization, and limiting forms

The half-length $2a$2 is the distance from the center to the fore and aft stagnation points $2a$3, and the half-height $2a$4 is the maximum half-width at $2a$5. The manuscript gives the useful relation

$2a$6

It further states that $2a$7 is determined from the $2a$8 boundary condition and that $2a$9, $Re=10$0, and $Re=10$1 are linked implicitly.

The aspect ratio is defined as

$Re=10$2

To compare shapes consistently, all ovals are rescaled so that

$Re=10$3

This normalization fixes the vertical diameter across the full family, so Reynolds number and force coefficients are evaluated on a common geometric basis.

Within this normalization, the geometry over $Re=10$4 follows a nearly linear trend. At $Re=10$5, the body is a circle with $Re=10$6, $Re=10$7, and $Re=10$8. At $Re=10$9, the body has $200$0, $200$1, and $200$2, i.e. an oval about three diameters long. The approximation

$200$3

is derived and used over the studied range (Xu et al., 8 Sep 2025).

$200$4	$200$5 trend	Geometric interpretation
$200$6	$200$7	Circular cylinder
$200$8	$200$9	Slender oval, about three diameters long

Two limiting statements organize the family. First, $Ua/m=0$0 corresponds to a circular cylinder. Second, increasing $Ua/m=0$1 produces progressively more elongated Rankine ovals. This makes the family particularly useful for studying shape effects continuously, without changing the underlying analytic construction.

3. Inviscidly streamlined, viscously bluff

In inviscid potential flow, the Rankine oval is “streamlined” because it is literally defined by a smooth streamline of an irrotational velocity field. In that setting there is no boundary layer, no separation, and no wake in the viscous sense. The pressure field follows directly from the potential solution, and the body is an analytically convenient idealization.

In viscous flow, by contrast, the same shape behaves as a bluff body. The no-slip boundary condition generates boundary layers, separation, recirculating wake bubbles, and eventually periodic vortex shedding. The central conceptual point is that streamline geometry in potential flow does not prevent wake instability once viscosity is included (Xu et al., 8 Sep 2025).

This establishes a bridge between two classical viewpoints. On one side is potential-flow theory, in which Rankine ovals are canonical exact shapes defined by source–sink superposition. On the other side is practical wake dynamics, in which the same bodies exhibit the standard separated-flow sequence seen behind cylinders, ellipses, rectangles, and other bluff bodies: steady recirculation at low $Ua/m=0$2, followed by a Hopf bifurcation to periodic shedding.

A common misconception is that a body derived from an inviscid streamline should remain dynamically streamlined in real flow. The reported results contradict that inference. A plausible implication is that analytic streamline constructions and viscous wake stability belong to different problem classes: the former constrains the outer inviscid geometry, whereas the latter is governed by no-slip-induced boundary layers and their instability.

4. Governing equations, simulation framework, and onset criterion

The viscous-flow study is based on two-dimensional direct numerical simulations of incompressible flow past Rankine ovals at zero angle of attack. The governing equations are the incompressible Navier–Stokes equations,

$Ua/m=0$3

$Ua/m=0$4

The simulations are 2D, so $Ua/m=0$5, corresponding to $Ua/m=0$6 and $Ua/m=0$7.

The solver uses a cell-centered finite volume method on hybrid unstructured meshes, a second-order Roe scheme for convection, a reconstructed central scheme for viscous flux, and second-order implicit dual-time stepping for unsteady simulations. The Reynolds number is defined by

$Ua/m=0$8

with $Ua/m=0$9 under the chosen normalization. The study also defines

$1$0

$1$1

and decomposes drag as

$1$2

The simulations cover $1$3 to $1$4 and $1$5 to $1$6, with finer sampling near shedding onset for each shape. The computational domain is

$1$7

with maximum blockage ratio $1$8. Boundary conditions are free-slip at the far field and no-slip on the Rankine oval surface. The first near-wall grid spacing is less than $1$9, with growth rate $Ua/m$0, and total cell count is about $Ua/m$1 to $Ua/m$2, depending on geometry (Xu et al., 8 Sep 2025).

Validation is performed against standard cylinder data at $Ua/m$3 and $Ua/m$4. For the circular-cylinder onset, the study obtains

$Ua/m$5

which is reported to match the literature well.

The onset of vortex shedding is identified using Stuart–Landau/Hopf-bifurcation analysis. The amplitude equation is

$Ua/m$6

where $Ua/m$7 is taken from the lift-coefficient envelope and $Ua/m$8. The onset criterion is based on the real growth rate $Ua/m$9, using

$+m$0

The critical Reynolds number is then determined by linearly fitting $+m$1 versus $+m$2 and locating $+m$3. This procedure directly ties the instability threshold to Hopf-bifurcation dynamics rather than to a purely visual wake criterion.

5. Wake instability, critical Reynolds number, and wake organization

The principal instability result is that the critical Reynolds number increases approximately linearly with $+m$4. The reported empirical fit is

$+m$5

stated to be reliable over

$+m$6

with minor deviation near $+m$7. Numerically, $+m$8 at $+m$9 and $(x,y)=(-a,0)$0 at $(x,y)=(-a,0)$1 (Xu et al., 8 Sep 2025).

The physical interpretation given is that increasing $(x,y)=(-a,0)$2 makes the body more slender and more streamlined geometrically, which tends to lengthen the recirculation region and stabilize the wake, so a higher $(x,y)=(-a,0)$3 is required to trigger the Hopf bifurcation. Fuller, more circle-like shapes therefore shed earlier.

Near $(x,y)=(-a,0)$4, the wake already exhibits the canonical anti-symmetric development associated with vortex shedding. At the phase of maximum lift, a single distinct vortex bubble is observed; when lift is zero, two wake vortex bubbles are visible. The asymmetry grows gradually as the wake approaches periodic shedding.

The study compares this trend with other two-dimensional bluff bodies. Rankine ovals and rectangles show roughly linear increases of $(x,y)=(-a,0)$5 with aspect ratio, whereas ellipses and diamonds show more nonlinear behavior. This suggests that Rankine ovals form an intermediate class: they are analytically generated by potential flow, yet their viscous instability thresholds vary with shape in a manner characteristic of separated bluff-body wakes.

The wake patterns are grouped into five classes based on vorticity intensity and shedding frequency. Increasing $(x,y)=(-a,0)$6 and decreasing $(x,y)=(-a,0)$7 both intensify vorticity and increase shedding frequency. Using rms velocity maps, the wake width $(x,y)=(-a,0)$8 is defined as the distance between the two rms peaks, and the vortex formation length $(x,y)=(-a,0)$9 as the streamwise distance from the body center to the midpoint between those peaks. At fixed $m$00, increasing $m$01 increases both $m$02 and $m$03; at fixed $m$04, increasing $m$05 decreases both. At $m$06, for example,

$m$07

At $m$08, $m$09 becomes nearly shape-independent, around $m$10–$m$11.

6. Forces, Strouhal scaling, and reduced-order drag prediction

At the critical point, increasing $m$12 decreases the critical mean drag coefficient from about $m$13 to $m$14 and increases the critical Strouhal number slightly from about $m$15 to $m$16. The approximate invariance

$m$17

is reported as consistent with previously observed bluff-body trends (Xu et al., 8 Sep 2025).

Over $m$18, the mean drag decreases with increasing Reynolds number for every shape, and at fixed $m$19 it decreases with increasing $m$20. The fit

$m$21

with

$m$22

implies

$m$23

so the friction contribution scales approximately like $m$24. The decomposition further shows that $m$25 is nearly constant with $m$26, whereas $m$27 decreases as $m$28 increases. At $m$29, for example, $m$30 over the studied range, while $m$31 drops from $m$32 to $m$33. For sufficiently elongated shapes, friction drag can become comparable to or exceed pressure drag; at $m$34 and $m$35,

$m$36

The instantaneous drag fluctuation is dominated by pressure-drag fluctuation, while friction drag remains comparatively steady. At $m$37, the amplitude of $m$38 is $m$39, compared with $m$40 for $m$41. As $m$42 increases, the amplitude of pressure-drag oscillation drops sharply, from $m$43 at $m$44 to $m$45 at $m$46. At fixed $m$47, increasing $m$48 reduces the amplitude of both lift and drag oscillations: the maximum instantaneous $m$49 falls from $m$50 at $m$51 to $m$52 at $m$53, and drag oscillation amplitude falls from $m$54 to $m$55.

The Strouhal number shows two distinct descriptions. On the one hand, in the raw geometric parameterization, $m$56 increases with $m$57 for a given shape and decreases with $m$58 at fixed $m$59; by $m$60, $m$61 for all shapes, indicating weaker geometry dependence in that regime. At $m$62,

$m$63

The study also notes that once $m$64 exceeds $m$65, $m$66 appears near $m$67 rather than growing gradually from zero.

On the other hand, the data-driven dimensional analysis identifies reduced variables in which shape dependence largely collapses. For friction drag, the selected variables are $m$68, with half-length-based coefficient

$m$69

The dimensional-analysis form is

$m$70

and the dominant controlling group is essentially

$m$71

The reported conclusion is that $m$72 is determined by $m$73 independently of $m$74, with fit

$m$75

For vortex shedding, the selected wake variables are the vortex formation length $m$76 and the largest reverse-flow speed $m$77, yielding

$m$78

The paper states that $m$79 is independent of $m$80 and controlled only by $m$81, with fitted relation

$m$82

The sign in the printed formula is identified in the source text as likely typographically problematic, because the plotted trend and accompanying discussion indicate a meaningful positive correlation over the sampled range. The explicit claim retained by the study is that the dependence collapses onto a single law in terms of $m$83, essentially independent of the shape parameter.

A further reduced-order result concerns pressure drag estimation. At $m$84, the time-averaged surface pressure distributions from DNS approach inviscid potential-flow predictions as $m$85 increases. This motivates the piecewise approximation

$m$86

for which the intended expression is explicitly reconstructed in the source text. The approximation works poorly for short, full bodies, with pressure-drag error exceeding $m$87 at $m$88 and still $m$89 at $m$90, but becomes accurate for sufficiently slender ovals; at $m$91, the estimated pressure drag is $m$92, within $m$93 of DNS (Xu et al., 8 Sep 2025).

Combining this pressure estimate with

$m$94

the total drag for sufficiently large $m$95 can be predicted without full numerical simulation by obtaining $m$96 from potential flow plus rear correction, computing $m$97 from $m$98, evaluating $m$99, obtaining $2a$00, converting to $2a$01, and then summing $2a$02. This suggests a limited but practically useful persistence of potential-flow structure within a viscous separated-flow regime.

The Rankine oval thus occupies a distinctive position among canonical bodies. It is an exact streamline body in inviscid theory, yet in viscous flow it falls squarely within the bluff-body instability framework. The contrast between its elegant potential-flow construction and its separated, shape-sensitive wake dynamics is the defining insight of the recent viscous-flow study.