CASE STUDY · 03 · PHYSICS-INFORMED NEURAL NETWORK

Multi-Output Aerodynamic Prediction with Physics-Informed Neural Networks

A physics-informed neural network that predicts an airfoil's lift, drag, and moment in a millisecond, trained on 52,480 NASA CFD simulations. The full network runs live below, in your browser, on TensorFlow.js. Drag the shape and watch the aerodynamics update in real time.

SCROLL TO ENTER
01  THE BOTTLENECK

CFD is the truth, but it is slow

Every wing, blade, and UAV begins with one question: for this shape, at this speed and angle, what are the lift, drag, and moment? Computational fluid dynamics answers it accurately by solving the governing equations on a mesh, but a single high-fidelity run can take hours.

Design optimization needs that answer thousands of times, across a whole space of shapes and flow conditions. At CFD speed, that loop is impractical. The goal of this project was a surrogate: a model that keeps CFD-level accuracy while returning an answer instantly, so a designer can explore the whole space interactively.

SURROGATE MODELAERODYNAMICSDESIGN OPTIMIZATION
CFD vs THIS MODEL
High-fidelity CFD
~ hours
This PINN
~ 1 ms
Once trained, the network predicts the force coefficients orders of magnitude faster than a full simulation, with no mesh and no solver.
READ MORE · METHODCONTEXT, GAP, AND METHOD

The paper asks a practical design question: can one surrogate remain accurate across geometry, Reynolds number, Mach number, and angle of attack, including stall and airfoils excluded from training? Its answer is a geometry-aware, physics-informed neural network trained on high-fidelity CFD.

01 · CFDPALMO flow solutions
02 · GEOMETRY18D CST encoding
03 · LEARNINGFDL + physics curriculum
04 · OUTPUTCl, Cd, Cm instantly

Why a new surrogate?

High-fidelity CFD resolves the governing flow equations and is the reference used here, but repeated meshing and solution make it too expensive for multidisciplinary design optimization. Conventional response surfaces and ordinary neural networks are fast, yet they can interpolate without learning the correct aerodynamic trends, particularly near sparse regions, stall, and a new geometry.

The related work reviewed in the paper spans data-driven MLP and CNN surrogates, physics-informed neural networks, generative airfoil design, and compact parameterizations such as CST. The unresolved issue is not only test-set accuracy; it is whether the model remains interpretable and physically plausible when the airfoil itself was never present during training.

That distinction changes the research target. A random train/test split can reward memorization because neighboring operating points from the same airfoil appear on both sides. The method therefore treats geometry as a first-class variable and measures success on entire shapes removed from optimization. It also predicts Cl, Cd, and Cm together because an efficient design cannot be judged by lift alone; drag determines efficiency and pitching moment determines trim and stability.

What this work contributes

Compact geometry

Every coordinate file is converted into upper and lower CST coefficients, giving the network a continuous shape description instead of an arbitrary cloud of points.

Derivative features

The Fast Derivative Layer exposes how learned features change with angle of attack, then fuses that sensitivity into the shared representation.

Physics curriculum

Data fit, derivative constraints, drag-polar structure, hard-case sampling, and tail losses are introduced in stages rather than optimized all at once.

How it fits into a design loop

A conventional optimization loop proposes a geometry, builds a mesh, runs the flow solver, extracts coefficients, and repeats. Here, CFD is paid for once when constructing PALMO. Each later candidate is normalized, encoded by CST, evaluated by the network, and screened in milliseconds. Promising candidates can still be returned to CFD for confirmation, turning the surrogate into a fast search layer rather than an unqualified replacement for simulation.

The paper's method is deliberately coefficient-focused. It does not reconstruct pressure or velocity fields, so it is far lighter than a field surrogate and directly matches the quantities required by many conceptual and preliminary design studies.

Core distinction: this is not a flow-field solver and it does not replace CFD outside its trained design family. It is a coefficient surrogate for fast exploration inside a documented aerodynamic envelope.

02  THE DATA · PALMO

52,480 CFD simulations from NASA

The model learns from PALMO, a high-fidelity dataset generated at NASA Ames with the OVERFLOW solver and Spalart-Allmaras turbulence closure.

It covers the NACA 4-series (from 0006 to 4424) across an unusually broad envelope: Mach 0.25 to 0.90, Reynolds 75k to 8M, and angle of attack −20° to +20°. Twelve airfoils were used for training; four were held out entirely as a strictly unseen test of generalization.

NASA PALMOOVERFLOW · SA16 NACA 4-SERIES
THE 16 NACA AIRFOILS · ★ = UNSEEN
The exact geometries in PALMO, drawn from their CST encodings. The four starred airfoils were withheld entirely from training, to test true generalization — each is still swept through every Mach × Reynolds × AoA condition.
READ MORE · DATA52,480 CFD CASES

PALMO is treated as the aerodynamic ground truth. It was produced at NASA Ames with OVERFLOW, using second-order temporal accuracy, fourth-order spatial differencing, and the Spalart-Allmaras turbulence model.

AXISVALUES USED IN THE PAPERROLE
Geometry16 NACA 4-series airfoils, 0006 through 4424Shape generalization
Mach0.25, 0.35, 0.45, 0.55, 0.65, 0.70, 0.75, 0.80, 0.85, 0.90Subsonic to transonic effects
Reynolds75k, 125k, 250k, 500k, 1M, 2M, 4M, 8MViscous-scale variation
AoA−20° to +20° at integer incrementsLinear range, stall, and reverse loading
TargetsLift Cl, drag Cd, pitching moment CmCoupled aerodynamic response

A split that actually tests a new shape

Twelve airfoils supply the development data. Their cases are divided 80/20 into training and held-out test samples. Four complete geometries, NACA 4421, 3418, 4415, and 3415, are removed before training and used only for the unseen-airfoil evaluation. This is stricter than randomly hiding operating points from a shape already known to the network.

Seen test

New Mach, Reynolds, and AoA combinations drawn from airfoil families represented in training.

Strictly unseen

All operating points belong to four geometries whose CST vectors were never used to fit the network.

Explore the simulation lattice

PALMO condition matrix
10 Mach levels × 8 Reynolds levels. Hover or tap a cell to inspect its contribution.
Each cell contains 16 airfoils × 41 angles = 656 CFD cases.

The lattice is balanced by construction: every geometry is sampled at the same flow conditions and integer angles. This makes cross-geometry comparisons cleaner, while the 41-point AoA sweep supplies enough ordered samples to learn the attached-flow slope, nonlinear transition, and post-stall behavior. Multiplying 16 × 10 × 8 × 41 gives the reported 52,480 simulations.

Deployment envelope: the paper trains on the complete −20° to +20° database. The browser demo intentionally presents −10° to +15°, the more useful interactive design range requested for this page. Reynolds and Mach retain the paper's full ranges.

03  GEOMETRY · CST

An airfoil in a handful of numbers

A neural network cannot ingest a raw list of surface points. The shape is encoded with the Class-Shape Transformation (CST): a class function fixes the round nose and sharp tail, and a small set of Bernstein weights bend the surface between them.

Eighteen coefficients describe an entire airfoil, smoothly and physically. That compactness is what lets the model generalize to shapes it has never seen. The animation shows the Bernstein basis blending into the final curve.

CSTBERNSTEIN BASIS18 COEFFICIENTS
LIVE · CST BASIS → SHAPE
Class function × Bernstein shape functions, summed with weights, reconstruct the upper and lower surfaces.
READ MORE · CST FITFROM COORDINATES TO 18 FEATURES

CST separates the universal topology of an airfoil from the coefficients that describe a particular shape. The same differentiable equation is used for the paper preprocessing and for the draggable browser geometry.

y(x) = C(x) S(x) + x Δz    |    C(x) = x0.5(1 − x)1.0    |    S(x) = Σ wi Bi7(x)

The class exponents create a rounded leading edge and a finite, sharp trailing edge. The shape function is an order-7 Bernstein expansion, so each surface has eight weights. Upper and lower trailing-edge offsets add two more inputs, producing the 18 geometry variables used by the model.

THE GEOMETRY INNOVATION The CST fit is explicitly coupled at the leading edge so the reconstructed upper and lower surfaces meet with a compatible tangent instead of forming two sharp, disconnected lines.

By tying the first upper coefficient to the sign-reversed first lower coefficient, the fit retains the rounded CST nose and a differentiable surface transition. This is the key idea behind making sparse coordinate files usable as smooth neural-network inputs without losing leading-edge physics.

Leading-edge continuity lab
Change the first CST weight, then compare the coupled parametric tangent with an unconstrained nose.
NOSE WEIGHT0.170
Coupled: wᵤ₀ = −wₗ₀, so both surfaces share one rounded nose constraint.

Coordinate preparation and robust fitting

  1. Move the leading edge to the origin, align the chord, normalize it to unit length, and identify the trailing edge.
  2. Split the coordinate loop into upper and lower surfaces and resample them consistently.
  3. Fit the two Bernstein expansions independently with robust least squares and a soft-L1 residual.
  4. Couple the first upper and lower coefficients with opposite sign, preserving leading-edge symmetry and tangent continuity.
  5. Reconstruct the profile and reject perturbations that create surface crossing, excessive thickness or camber, or an invalid nose and trailing edge.

The robust objective is important near the leading and trailing edges, where coordinate ordering, duplicate points, and small measurement offsets can otherwise dominate an ordinary least-squares fit. Soft-L1 behaves quadratically for normal residuals but reduces the influence of isolated bad points. After fitting, reconstruction error and geometric validity are checked in physical coordinates before the coefficients enter the learning dataset.

8 upper

Bernstein weights describe suction-side curvature from nose to trailing edge.

8 lower

A separate expansion represents the pressure side without forcing symmetry.

2 offsets

Upper and lower Δz terms retain the measured trailing-edge closure.

Why the live shape now looks correct: the visible curve is densely evaluated from the CST equation, not connected through a few draggable points. Handles only perturb nearby valid CST coefficients, while the reconstructed surface remains smooth at the nose and along the chord.

04  THE NETWORK

Physics, built into the wiring

The inputs — Reynolds, Mach, angle of attack, and the CST geometry — pass through a shared encoder, then split into three heads for lift, drag, and moment.

Two ideas make it physics-informed. The angle of attack gets a frequency encoding so the network can represent sharp, periodic effects like stall. And a Fast Derivative Layer computes ∂(hidden)/∂α and injects those sensitivities back into the features, so physical reasoning lives inside the network, not just in the loss. That one change cut the error on unseen airfoils by nearly ten times.

FAST DERIVATIVE LAYERFREQUENCY ENCODINGMULTI-TASK HEADS
ARCHITECTURE
Shared encoder → Fast Derivative Layer (∂/∂α) → three regression heads. Signed-log transforms keep the coefficients well-scaled.
READ MORE · NETWORK23 INPUTS, FDL, 3 HEADS

The model combines physical state, compact geometry, a frequency representation of angle of attack, and learned angle sensitivities. Three specialized output heads then share the same aerodynamic representation.

Complete architecture diagram from physical and CST inputs through the encoder and Fast Derivative Layer to lift, drag, and moment heads
FIGURE · COMPLETE ARCHITECTURE FROM THE PAPER. THE RAW 23D INPUT IS AUGMENTED BY SIX AOA FREQUENCY BANDS BEFORE THE SHARED ENCODER.

From 23 physical values to three coefficients

STAGEDIMENSION / OPERATIONPURPOSE
Raw inputln(Re), AoA, Mach + 18 CST geometry values = 23One vector describes both flow and shape.
AoA encoding6 bands, 12 sine/cosine channels; 23 → 35Resolves nonlinear and rapidly changing angular behavior.
Encoder35 → 384 → 192, SiLU + dropoutLearns a shared aerodynamic latent state.
FDLFirst 48 hidden sensitivities ∂h/∂α; 192 + 48 → 240Makes angle response available as a feature.
HeadsThree branches, 240 → 192 → 96 → 1Separate Cl, Cd, and Cm specialization.

Fast Derivative Layer

During gradient-enabled training, automatic differentiation calculates the derivative of the first 48 latent features with respect to AoA. The derivatives are scaled by 0.05 and concatenated with the 192-dimensional encoder output. This differs from a standard PINN that uses derivatives only as a penalty: here the sensitivity becomes part of the representation consumed by every output head.

For exported inference, the derivative branch is zero-filled, matching the original notebook implementation. Its training influence is still embedded in the learned encoder and head weights. This detail matters because the browser must reproduce the saved model, not silently invent a different inference graph.

FDL generalization ablation
Relative RMSE gap, unseen airfoils versus the held-out test set. Lower is better.
FDL reduces the reported unseen/test relative-RMSE ratio from 2.9495 to 0.3195.
Ablation plots comparing models with and without the Fast Derivative Layer
ABLATION · REMOVING FDL GREATLY WIDENS THE GENERALIZATION GAP. THE REPORTED UNSEEN-TO-TEST RRMSE RATIO RISES FROM ABOUT 0.32 TO 2.95 IN THE PAPER'S COMPARISON.

Stable targets and hard cases

Cl, Cm:   y′ = sign(y) log(1 + |y|)    |    Cd:   y′ = log(1 + Cd)

Signed-log transforms preserve the direction of lift and moment while compressing extremes. Drag uses log1p because it is nonnegative, and the inverse path applies expm1 plus a nonnegative clamp. StandardScaler statistics are learned from training data only. A weighted sampler later gives higher probability to examples whose error exceeds the current mean, concentrating capacity on stall and other difficult regions.

The six-frequency AoA encoding is concatenated with, rather than substituted for, the physical angle. The raw coordinate preserves ordering and extrapolation direction, while sine and cosine channels give the MLP a richer basis for steep curvature near stall. The three heads share low-level flow and geometry information but can specialize their final layers for the very different scales and shapes of Cl, Cd, and Cm.

Signed logarithmic output transform and inverse mapping used for aerodynamic coefficients
PREPROCESSING · REVERSIBLE TARGET TRANSFORMS REDUCE SCALE IMBALANCE WITHOUT LOSING THE SIGN OF CL OR CM.
05  TRAINING · PHYSICS LOSS

Taught the rules, not just the data

Fitting the data is not enough; the predictions must obey physics. Training ran as a multi-phase curriculum that slowly turns up physics-based penalties.

Three constraints are enforced softly: lift should rise with angle before stall (∂Cl/∂α > 0), the moment should keep the airfoil stable (∂Cm/∂α < 0), and drag must follow a proper polar (Cd ≈ Cd₀ + k·Cl²) and never go negative. This is what keeps the surrogate trustworthy where data is thin.

CURRICULUMDRAG POLARSTABILITY
THE PHYSICS CONSTRAINTS
  • Lift slope — ∂Cl/∂α > 0 before stall
  • Stability — ∂Cm/∂α < 0
  • Drag polar — Cd ≈ Cd₀ + k·Cl²
  • Non-negative drag — Cd ≥ 0 always
  • Hard-example mining — focus on the cases it gets wrong
Physics enters as extra loss terms, ramped up over three phases so the network learns the data first, then the laws.
READ MORE · TRAININGDATA, PHYSICS, TAILS

The objective is a controlled mixture of data accuracy and aerodynamic structure. Masks and gates restrict each physics rule to the regime where it is valid, so the optimizer does not force pre-stall behavior onto stalled or compressibility-dominated samples.

Diagram of supervised, relative, coverage, tail, ratio, and physics losses used in training
LOSS SYSTEM · MULTIPLE TERMS SHARE ONE PURPOSE: ACCURATE COEFFICIENTS THAT KEEP THE RIGHT DERIVATIVES, SIGNS, AND COUPLED TRENDS.

Physics terms

Pre-stall lift

Penalizes negative ∂Cl/∂α in the attached-flow window, with stall and Mach gating.

Static stability

Discourages positive ∂Cm/∂α where a stable pitching response is expected.

Drag polar

Encourages Cd = Cd₀ + kCl² locally, rather than fitting drag as an unrelated scalar.

Physical bounds

Prevents negative drag and suppresses implausible behavior using flow and geometry masks.

Data and robustness terms

The supervised core uses SmoothL1, which behaves quadratically around small errors and linearly for large residuals. Relative-error shaping balances coefficients with different scales. Coverage terms discourage systematic misses, a tail loss protects extreme cases, and a Cl/Cd ratio term improves the derived efficiency metric. The weighted hard-example sampler is activated only after the base map has stabilized.

Each rule is evaluated through a validity mask. For example, positive lift slope is a useful attached-flow prior but becomes incorrect beyond stall, and a simple quadratic drag polar is most defensible away from strong transonic effects. Mach, AoA, stall, and geometry gates therefore decide where a penalty contributes. This keeps “physics-informed” from becoming “physics-forced” in regimes where the simplified rule no longer applies.

PHASEFOCUSWHAT CHANGES
1 · FoundationData fit + physicsPhysics weight 0.5; no tail loss, ratio polishing, or hard sampler.
2 · Hard casesPhysics + tailsPhysics weight 0.4; hard-example mining and tail protection are enabled.
3 · PolishRelative and efficiency accuracyPhysics weight 0.2; stronger Cl/Cd ratio and relative-error optimization.
Curriculum weight explorer
Select a phase to see which objectives dominate and why.
Phase 1 establishes the supervised CFD map with the strongest physics regularization.

Training runs for 220 epochs with learning-rate warmup followed by cosine decay, checkpointing the best validation model. On an RTX 3090 the reported training time is under 30 minutes.

Training validation and unseen loss curves across the multi-phase curriculum
CURRICULUM ABLATION · TRAIN, TEST, AND UNSEEN CURVES REMAIN CLOSE ACROSS PHASE TRANSITIONS.
Normalized Cl over Cd error comparison for single-phase and three-phase training
EFFICIENCY ABLATION · THREE-PHASE TRAINING SUBSTANTIALLY LOWERS NORMALIZED CL/CD ERROR ON TEST AND UNSEEN CASES.

Why the order matters: enforcing every objective from epoch one makes gradients compete before the network has learned the basic CFD map. The staged schedule first establishes that map, then concentrates on physics, tails, and ratios.

06  RESULTS

CFD accuracy, at interactive speed

On the held-out test set the model reaches R² = 0.9997 for lift, 0.9991 for drag, and 0.9990 for moment.

The real test is the four airfoils it never saw during training. There it stays strong (R² ≈ 0.999 for lift and drag), which is exactly what the Fast Derivative Layer and the physics curriculum were built to protect. A prediction that used to cost hours of CFD now costs about a millisecond.

R² ≈ 0.999UNSEEN-ROBUST~1 ms
R² · TEST vs UNSEEN
Accuracy holds from familiar airfoils to strictly-unseen geometries, the hallmark of a model that learned the physics, not just the examples.
READ MORE · EVIDENCEEVIDENCE FROM THE PAPER

The evaluation checks more than one aggregate score: coefficient parity, complete angle sweeps, three-dimensional operating maps, ablations, an external surrogate comparison, symmetry, and a deliberately difficult out-of-family experiment.

SPLITCLCDCM
Held-out test R²0.99970.99910.9990
Held-out test RMSE0.01230.00430.0033
Unseen-airfoil R²0.99910.99920.9968
Unseen-airfoil RMSE0.01780.00400.0045
Accuracy explorer
Switch split and metric. R² uses a focused 0.99–1.00 scale; RMSE starts at zero.
Held-out test R²: Cl 0.9997, Cd 0.9991, Cm 0.9990.

R² measures how much variation in the CFD outputs is explained, while RMSE preserves each coefficient's physical scale. Both are needed: a high R² can coexist with a practically important local error, and direct RMSE values are not comparable across coefficients without remembering that drag and moment are numerically smaller than lift.

Predicted versus CFD parity plots for lift drag and moment on test and unseen airfoils
PARITY · PREDICTIONS TRACK THE CFD DIAGONAL FOR CL, CD, AND CM ON BOTH SEEN-FAMILY TEST DATA AND STRICTLY UNSEEN GEOMETRIES.

Does it reproduce complete aerodynamic curves?

For the unseen NACA 3415 at Mach 0.25 and Re = 1M, the PINN follows PALMO across the full angle sweep, including the nonlinear transition and stall near −10°. The paper's NASA surrogate/interpolation baseline misses this behavior more clearly, illustrating why global R² alone is not sufficient.

Curve-level agreement tests continuity and ordering, not only isolated residuals. The lift curve should recover the linear slope and both stall branches; drag should stay positive and rise as loading grows; Cm should maintain a plausible stability trend; and Cl/Cd should peak in a localized efficient corridor rather than oscillate between adjacent angles.

Comparison of PINN predictions, PALMO CFD, and NASA surrogate for NACA 3415
BASELINE COMPARISON · PINN, PALMO CFD, AND THE NASA SURROGATE ON NACA 3415.
Lift drag moment and efficiency curves for unseen NACA 3415
ANGLE SWEEP · CL, CD, CM, AND CL/CD FOR A STRICTLY UNSEEN AIRFOIL.

Maps over Mach and angle of attack

The surface and contour plots show that the network learns coupled aerodynamic structure rather than isolated point predictions. Lift grows through the attached-flow range and rolls off around stall; drag rises rapidly at high loading and compressibility; moment changes consistently with camber and stability; efficiency peaks in a bounded operating corridor. The symmetric NACA 0012 case also provides a useful reversal and near-zero-moment check.

Predicted aerodynamic response surfaces for unseen NACA 3415 across Mach and angle of attack
NACA 3415 · STRICTLY UNSEEN RESPONSE SURFACES.
Predicted aerodynamic response surfaces for unseen NACA 4421
NACA 4421 · SECOND UNSEEN-GEOMETRY STUDY.
Aerodynamic response surfaces for symmetric NACA 0012
NACA 0012 · SYMMETRIC-AIRFOIL CHECK FOR LIFT REVERSAL, MOMENT, DRAG, AND EFFICIENCY.

What each ablation established

  • FDL: removing hidden angle sensitivities sharply damages unseen-shape robustness.
  • Three-phase curriculum: a single training phase produces a larger train-to-unseen gap and worse efficiency prediction.
  • Signed-log and frequency features: target transforms and AoA encoding reduce drag error, especially at extreme angles.
  • Physics losses: derivative and drag-polar constraints improve trends in sparse regions instead of merely improving pointwise averages.

Limitations and the failure study

The model is intentionally scoped to the NACA 4-series distribution represented in PALMO. To test a more difficult geometric gap, the paper removes NACA 2406 and close variants from training. The network still identifies the broad stall pattern, but peak Cl and Cl/Cd become less accurate. That result sets an important boundary: physics guidance improves interpolation and nearby generalization, but it cannot replace representative CFD coverage.

This failure case is scientifically useful because it separates smooth response regularization from geometric knowledge. The derivative losses can teach how coefficients ought to vary with AoA, but they cannot infer an unobserved thin-airfoil response perfectly when nearby thickness and camber combinations are absent. A production system should therefore measure distance in standardized CST space, flag remote candidates, and route the most informative ones back to CFD.

Airfoils removed in the out-of-family generalization experiment
STRESS TEST · NACA 2406 AND NEARBY SHAPES REMOVED TO CREATE A GEOMETRY GAP.
Reduced peak accuracy for NACA 2406 after nearby airfoils are removed from training
LIMIT · STALL LOCATION REMAINS ROBUST, WHILE PEAK CL AND CL/CD LOSE ACCURACY.

Next steps from the paper: expand beyond NACA 4-series families, quantify predictive uncertainty, and use targeted data acquisition to add CFD cases where uncertainty or geometric distance is highest.

07  TRY IT · LIVE IN YOUR BROWSER

Design an airfoil, feel the aerodynamics

The exact trained network, running on TensorFlow.js. Drag the CST geometry, set the flow conditions, and compare the live prediction against PALMO CFD whenever the selected case exists in the training database.

READ MORE · LIVE MODELREAL WEIGHTS, CST, CFD OVERLAY

This is the exported trained surrogate, not a hand-written aerodynamic approximation. The page reconstructs the PyTorch layers and exact float32 checkpoint weights in TensorFlow.js, then applies the paper's feature order, scalers, frequency encoding, output transforms, and inverse transforms.

Geometry path

Preset coordinates are represented by the same 18D CST form. Dragging changes bounded CST parameters and reruns the validity guard before prediction.

Flow path

Mach 0.25–0.90, Reynolds 75k–8M, and AoA −10° to +15° stay inside the intended browser envelope.

Evidence path

For an unchanged database airfoil, the charts draw PALMO CFD beside the prediction. Once geometry is edited, only the model curve is shown.

Exact inference sequence

  1. Validate the edited CST vector and reconstruct a dense, non-intersecting airfoil surface.
  2. Build the 23 physical inputs in the notebook's original order, using ln(Re) and the saved training scalers.
  3. Append all 12 sine/cosine AoA features and evaluate the shared encoder plus the three trained heads.
  4. Undo standardization and signed-log transforms to recover physical Cl, Cd, and Cm.
  5. Repeat across the displayed AoA grid, then retrieve a PALMO reference only when geometry and flow conditions match a stored case.

Reading the plots correctly

The lift and efficiency panels are evaluated over every half degree in the displayed AoA range. The drag polar uses the same predictions with Cd on the horizontal axis and Cl vertically. Smooth strokes are only a visual interpolation between deterministic samples; the coefficient values are not randomized or filtered with synthetic noise.

When a preset exactly matches a PALMO geometry and database condition, the reference curve provides a direct CFD comparison. For a custom CST shape there is no CFD record to display, so the page labels it as prediction-only. Geometry guards limit thickness, camber, control-point movement, surface crossing, nose continuity, and trailing-edge closure, reducing extrapolation into obviously invalid airfoils.

The current white marker is the selected AoA, while the full line is a fresh batch evaluation at the chosen Mach, Reynolds number, and geometry. This makes local coefficient changes and global curve changes visible together. Because all output values come from one deterministic forward pass, moving a control does not inject measurement noise or randomly jitter the plot.

Scientific boundary: an interactive prediction is reliable only to the extent that its geometry and operating condition remain close to the PALMO/NACA 4-series domain. The demo exposes that boundary rather than presenting arbitrary shapes as CFD truth.

AIRFOIL · DRAG THE POINTS
PRESET
MACH0.45
Re1.00 × 10⁶
AoA °4.0
LIFT · Cl
DRAG · Cd
MOMENT · Cm
EFFICIENCY · Cl/Cd
LIFT · Cl vs AoA
EFFICIENCY · Cl/Cd vs AoA
DRAG POLAR · Cl vs Cd
Loading the network…
·  THE TAKEAWAY

A wind tunnel that fits in a web page

Compact geometry, a physics-aware network, and a curriculum that teaches the laws add up to a surrogate you can actually design with: CFD-level answers, instantly, even on shapes it never trained on. Everything above ran the real weights, live.

READ MORE · CONCLUSION

The paper's central result is that compact CST geometry, latent angle derivatives, and staged physics-aware optimization can preserve near-CFD coefficient accuracy while cutting inference to interactive time. The unseen-airfoil tests show strong interpolation across the studied NACA family; the NACA 2406 stress test shows why training-envelope coverage still matters.

The proposed research direction is therefore not to abandon CFD, but to use it strategically: broaden the geometry families, attach uncertainty estimates to each prediction, and request new high-fidelity cases only where the surrogate is uncertain. That closes the loop between accurate simulation and rapid design exploration.