Adjoint 2D Boundary Layer Equations

• Adjoint two-dimensional boundary layer equations are derived from the classical Prandtl formulation to quantify how perturbations affect drag and wall shear.
• They employ similarity transformations, spectral expansions, and Libby–Fox perturbation theory to yield explicit sensitivity kernels and analytical adjoint solutions.
• These equations underpin optimal flow control and shape optimization in laminar viscous flows, demonstrating accurate predictive capabilities in drag reduction strategies.

Adjoint two-dimensional boundary layer equations provide the mathematical foundation for sensitivity analysis, optimal flow control, and shape optimization in laminar viscous flows over surfaces. Arising as the formal adjoints of the classical Prandtl or boundary-layer equations, these systems describe how perturbations or objective functionals (such as drag or wall shear) respond to infinitesimal changes in parameters or boundary conditions. The analytic structure and physical interpretation of adjoint boundary-layer equations, particularly for the canonical Blasius (flat plate, zero-pressure-gradient) flow, have been rigorously established and linked to classical perturbation theory (Libby–Fox modes), providing spectral representations, Green’s functions, and explicit sensitivity kernels for drag-modification strategies (Kühl et al., 2020, Lozano et al., 23 Jan 2026).

1. Governing Equations and Adjoint Construction

The two-dimensional, steady, incompressible boundary-layer equations for flow $(u,v)$ over a flat plate are

$$\begin{aligned} &u\,\frac{\partial u}{\partial x} + v\,\frac{\partial u}{\partial y} = \nu\,\frac{\partial^2 u}{\partial y^2}, \ &\frac{\partial u}{\partial x} + \frac{\partial v}{\partial y} = 0, \end{aligned}$$

where $\nu$ is the kinematic viscosity. The wall shear stress $\tau_w(x)$ and integrated skin-friction coefficient $C_f(L)$ are natural objective functionals: $$C_f(L) = \int_0^L \tau_w(x) dx, \quad \tau_w(x) = \mu\,u_y(x,0).$$

The adjoint system is formally derived by introducing Lagrange multipliers $(w, q)$ and variational calculus. In velocity–pressure form, the continuous adjoint boundary-layer equations are

$$\begin{aligned} u\,w_x + v\,w_y - \nu\,w_{yy} - u\,q_y + v\,q_x = 0,\ q_y - w_x = 0, \end{aligned}$$

with boundary conditions $w(x,0)=0$, $w(L,y)=0$, $w(x,\infty)=0$, $q(x,0)=2\,\tau_w(x)$ (Lozano et al., 23 Jan 2026). For the generalized system including ATC (adjoint transpose convection) fluxes and arbitrary functionals $j_\Omega$, the adjoint BL system reads (Kühl et al., 2020): $$\begin{aligned} &-v_1\,\partial_{x_1}\hat v_1 - v_2\,\partial_{x_2}\hat v_1 + \hat v_1\,\partial_{x_1}v_1 + \partial_{x_1}\hat p - \nu\,\partial_{x_2x_2}^2\hat v_1 = -\partial j_\Omega/\partial v_1,\ &\hat v_1\,\partial_{x_2}v_1 + \partial_{x_2}\hat p - \nu\,\partial^2_{x_2x_2}\hat v_2 = -\partial j_\Omega/\partial v_2,\ &-\frac{1}{\rho}(\partial_{x_1}\hat v_1+\partial_{x_2}\hat v_2) = 0. \end{aligned}$$

2. Similarity Reduction and Adjoint Blasius System

Exploiting self-similarity, the Blasius similarity coordinate

$\eta = y \sqrt{\frac{U}{\nu x}}$

reduces the PDEs to ODE form. The primal streamfunction is $\psi(x,y) = \sqrt{\nu\,U\,x}\,F_0(\eta)$, yielding the classical Blasius ODE: $$F_0''' + F_0\,F_0'' = 0, \quad F_0(0)=F_0'(0)=0,\; F_0'(\infty)=1.$$

The adjoint equations, reduced by the same similarity transformation, produce a pair of ODEs for the adjoint streamfunction $\hat f(\eta)$, for which the tangential component reads: $-2\hat f''' + f\hat f'' = \eta f''\hat f',$ with complementary conditions $$\hat f(0) = 0, \; \hat f'(0)=0, \;\hat f'(\infty)=1$$ (Kühl et al., 2020). The normal equation $f''\,\hat f'=0$ enforces, for nontrivial Blasius profiles, $\hat f'(\eta)=0$ unless $f''(\eta)\equiv 0$, thus simplifying the tangential ODE to

$-2\hat f''' + f\hat f'' = 0.$

3. Libby–Fox Perturbation Theory and Spectral Expansion

The Libby–Fox framework analyzes algebraic perturbations around the Blasius solution, yielding spectral decompositions of the adjoint field. Small-amplitude disturbances are represented as separated variables: $\delta F(x,\eta) = x^{-\sigma/2} N(\eta),$ and the linearized perturbation ODE,

$$N''' + F_0 N'' + F_0'' N' + 2(F_0' N'' - F_0'' N) = 0,$$

defines a Sturm–Liouville eigenproblem with discrete spectrum $\{\sigma_k\}$, eigenfunctions $N_k(\eta)$, and orthogonality relations with respect to the weight $\omega(\eta) = F_0''/F_0'$ (Lozano et al., 23 Jan 2026).

The analytic adjoint solution is constructed from the Green’s function of the Libby–Fox operator: $$G(x,\eta; x_0,\eta_0) = -H(x-x_0) \sum_{k=1}^\infty \frac{N_k(\eta) N_k'(\eta_0)}{C_k F_0'(\eta_0)} \left(\frac{x}{x_0}\right)^{-\sigma_k/2},$$ leading to the explicit mode-sum expansion for the adjoint wall shear,

$w(x,\eta) = -\frac{2}{\Rey\,L}\sum_{k=1}^\infty D_k(\eta) \left[\left(\frac{x}{L}\right)^{-(\sigma_k-2)/2} - \left(\frac{x}{L}\right)^{-\sigma_k/2}\right],$

where $D_k(\eta) = \frac{1}{C_k} N'_k(\eta)/F_0'(\eta)$ (Lozano et al., 23 Jan 2026).

4. Analytical Properties and Spectral Constraints

The eigenvalue quantization enforced by wall and infinity boundary conditions yields a discrete, positive spectrum $\{\sigma_k\}$. Orthogonality and normalization are governed by

$$C_k = \int_0^\infty \omega(\eta) N_k^2(\eta) d\eta, \qquad \int_0^\infty \omega N_k N_j d\eta = C_k \delta_{kj}.$$

A sum-rule constrains the spectral data: $$\sum_{k=1}^\infty \frac{(2-\sigma_k)^2}{C_k} = 1, \qquad \sum_{k=1}^\infty D_k(\eta) = 1,$$ guaranteeing completeness and consistency of the adjoint expansion (Lozano et al., 23 Jan 2026).

In the self-similar Blasius case, the so-called ATC (adjoint transpose convection) terms in the adjoint equations are found to cancel identically, rendering the adjoint and primal solutions coincident, i.e., $\hat f'(\eta) = f'(\eta)$ (Kühl et al., 2020). Consequently, explicit analytical expressions are available for the adjoint boundary layer thickness, wall shear, and skin-friction coefficients. For example,

$$\hat{\delta}_{99} = 5.9424 \sqrt{\frac{\nu x}{U}}, \quad \delta_{99} = 4.91 \sqrt{\frac{\nu x}{U}},$$

$$C_{f,\mathrm{adj}} = \frac{2\hat f''(0)}{\sqrt{\Re_x}} \approx \frac{0.369}{\sqrt{\Re_x}},$$

$$C_{d,\mathrm{adj}} \approx \frac{0.738}{\sqrt{\Re_L}}.$$

5. Shape Sensitivity and Applications in Flow Control

The adjoint field provides the kernel by which sensitivities with respect to boundary or volume perturbations are computed. The shape derivative of the integrated drag objective $J$ due to a wall-normal deformation is

$$\delta_u J = \int_{\Gamma_w} s\, d\Gamma, \qquad s = -\nu\,\left.\frac{\partial v_i}{\partial x_j}\right|_{w}\left.\frac{\partial \hat v_i}{\partial x_k}\right|_{w}\, n_j n_k,$$

with explicit expressions in Blasius coordinates (Kühl et al., 2020). For variations in wall-normal velocity (blowing or suction), the drag sensitivity becomes

$\delta C_f = \int w_y(x,0)\; v_w(x)\; dx,$

where $w_y(x,0)$ is obtained analytically from the spectral expansion (Lozano et al., 23 Jan 2026).

Such expressions directly enable shape optimization, control design, and drag reduction strategies in laminar boundary layers by providing precise analytic kernels for sensitivity.

6. Extensions to Falkner–Skan Flows

For outer flows with a power-law profile, $U(x) \sim x^m$, the Falkner–Skan similarity transformation generalizes the approach. The base flow satisfies

$$F''' + F F'' + \beta(1 - F'^2) = 0, \quad \beta = \frac{2m}{m+1},$$

while perturbations $N(\eta)$ solve a parameterized eigenproblem: $$N''' + F N'' + [2\beta + \sigma\beta -1] N' + 2(1 - \sigma - \beta) F'' N = 0,$$ again admitting discrete spectra and adjoint mode expansions. The Green's function and adjoint sensitivities adopt modified power-law forms analogous to the Blasius case (Lozano et al., 23 Jan 2026).

7. Predictive Validation and Physical Interpretation

Direct comparison with full Navier–Stokes and discrete adjoint simulations at $10^3 \le \Re_L \le 10^5$ shows the analytic, ATC-free adjoint theory predicts drag coefficients and thicknesses accurately (agreement within a few percent for $C_d$, $\delta_{99}$). Scaling of adjoint drag coefficients as $1/\sqrt{\Re_L}$ and sensitivities as $\Re_L^{-1}$ to $\Re_L^{-1/2}$ is observed (Kühl et al., 2020). The adjoint solution represents the backward-propagation of disturbances from objectives (e.g., plate trailing edge), which, for self-similar boundary layers, collapses onto the same similarity manifold as the primal flow. This symmetry permits analytic closed-form expressions, streamlining shape and flow control analyses in laminar boundary layers.