Airfoil design
Airfoil Optimization Explained: Improving Lift, Drag and Performance
HENDOI · 27 October 2026
Optimization changes the shape on purpose: variables, an objective, constraints, and a method. A lower drag number is not a result unless the constraints survived.
What you are allowed to change
An optimizer needs a short list of numbers that generate a shape. Those design variables might be the camber and thickness of a NACA 4-digit section (only two or three knobs, a small search), or the coefficients of a CST or PARSEC parameterization (more knobs, shapes the 1930s catalog never drew). More variables can represent a better section. They can also represent a wavy, unbuildable section that exploits a hole in the aerodynamic model. The geometry gate after the optimizer is part of the optimizer.
What you are trying to improve
An objective is a single number to push down or up. Common airfoil objectives:
- Minimize CD at a fixed CL, fixed Re and Mach. This is the cruise-point problem.
- Maximize L/D at that point. Equivalent only if CL is held fixed. If CL is free, the optimizer may wander off to a CL you do not fly.
- Maximize CL max, or a surrogate for it. Stall objectives are the ones viscous panel methods are weakest at. Treat the answer as a candidate, not a promise.
- A weighted sum of cruise drag and a moment penalty. Honest only if you can explain the weights in airplane units, not as numbers that made the plot look nice.
If you care about drag and moment and stall at once, you do not have one objective. You have a multi-objective problem. That lesson is later in this series. Pretending a weighted sum is “the” optimum hides the trades.
Constraints are the design
| Constraint | Why it exists |
|---|---|
| Thickness at the spar, or maximum thickness | The section must close around structure, fuel or a battery. |
| Leading-edge radius or a minimum thickness near the nose | Stall behaviour and manufacturing. An infinitely sharp nose is a numerical cheat. |
| CM not more nose-down than a limit | Tail size and trim drag. |
| Area or a specified CL | Stops the optimizer lowering drag by refusing to make lift. |
| Smoothness, curvature limits, trailing-edge angle | Stops oscillatory shapes that a panel code scores as low drag and a machinist cannot cut. |
Methods, at the level that changes your judgement
A gradient method follows the slope of the objective. It is fast when the analysis is smooth and the start point is close. It stops at a local optimum: the best nearby shape, not the best shape in the world. A global method (a genetic algorithm, a particle swarm, a well-run surrogate model) samples more widely and costs more evaluations. It can still miss. A surrogate — a response surface fitted to a few dozen analyses — is only as good as the samples and the aerodynamic model underneath.
The aerodynamic model is the quiet constraint. If the polar method cannot see stall, an optimizer that “maximizes CL max” is optimizing the method’s blind spot. If the method loves laminar flow, the optimizer will draw a section that is brilliant at Ncrit = 12 and ordinary with bugs. Re-evaluate the winner with a trip, at a second Reynolds number, and with a geometry check a human can plot. Then, if the decision is expensive, re-evaluate it with a higher-fidelity method.
A usable loop
- Write the mission CL, Re, Mach and the constraints in one paragraph before you run anything.
- Parameterize the shape with as few variables as the mission needs.
- Optimize.
- Plot the winner over the baseline. If you cannot see a physical reason for the drag change in the pressure distribution, do not accept it.
- Perturb Re, Ncrit and the constraint limits. If the winner disappears, it was never a winner.
HENFOIL’s optimize path is this loop with method labels kept on the evaluations. A lower CD from that path is a lower CD of a VIC evaluation under stated constraints. That is already useful. It is not a flight-test claim.
Where this sits in the series
In simple words: what this page is for
Optimization changes the shape on purpose: variables, an objective, constraints, and a method. A lower drag number is not a result unless the constraints survived.
This lesson exists because section decisions fail when geometry, method labels and selection reasons are scattered. Airfoil Optimization Explained: Improving Lift, Drag and Performance helps you keep a clean trail from shape to polar to choice.
Why the topic matters
Engineers lose time when a title sounds clear but the page only sketches the idea. This article expands Airfoil Optimization Explained: Improving Lift, Drag and Performance so you can answer three questions without jargon: what is the idea, when do you use it, and how do you check it in a HENDOI product.
Read the sections above for the core definitions. Use the steps below to move from reading into a real project trail you can share with a teammate.
How to apply this in HENDOI
In practice: create a HENDOI account or log in, open HENFOIL, bring a section in (create, import DAT, or digitize), pass the geometry gate, climb the analysis ladder only as far as the decision needs, then compare and archive. Every calculated number should keep its method label.
When the section choice is ready for the aircraft, continue in HENSYNTH for mission, compliance and trades. Same account. Same plans.
What this page will not claim: a teaching polar or geometry note is not a flight-Re certificate. HENFOIL labels methods so you know what you bought. Tunnel data and professional judgement still sit outside any single software run.
Checklist before you leave this page
- Can you state the idea of Airfoil Optimization Explained: Improving Lift, Drag and Performance in one sentence?
- Do you know which product owns the next action — HENFOIL or HENSYNTH?
- Will every number you keep carry a method label or a stored-run source?
- If you are new here, have you opened Register or Login?
Related reading on this site
- Active Flow Control on Airfoils: Suction, Blowing and Synthetic Jets
- What Is the Aerodynamic Center of an Airfoil? Location, Meaning and Applications
- How to Identify Numerical Convergence Problems in Airfoil Analysis
- How to Build an Airfoil Analysis Database with Consistent Simulation Conditions
- All HENDOI blog articles
- HENFOIL product page
- HENSYNTH product page
- Pricing — Student, Professional & Team
- FAQ
- Login · Register now
- Previous: How to choose an airfoil for a fixed-wing UAV
- Next: How to design an airfoil for long-endurance UAVs
- Index: all HENFOIL airfoil lessons
These pages teach the ideas. They are not wind-tunnel data and not a certification analysis. When a number matters, compute it at a stated Reynolds number, Mach number and transition setting, and keep the method label on the result. In HENFOIL that production polar is a VIC polar — see the analysis ladder and the polar field guide.
Both products live
Use this lesson in the HENDOI workspace
This page sits in Airfoil design. HENFOIL and HENSYNTH are both live on one account. These articles are screening lessons — not certification reports.
Live · section
HENFOIL
Turn this lesson into a section you can draw, check, analyse and compare. Run the polar at the Reynolds number of the vehicle. The method name stays on the result.
To Know MoreLive · aircraft
HENSYNTH
After the section is chosen, set geometry and mission requirements, run screening, read compliance, synthesize candidates and download or mail the report.
To Know MoreAerospace industry collaboration
Share thoughts and ideas — help us build better solutions
We are looking for aerospace industry collaboration. Share your thoughts, ideas and real workflow challenges so we can improve our products and deliver the best solutions for OEMs, UAV / eVTOL teams, labs and aeronautical engineering companies.
Written by

Related articles
- How to Choose the Right Airfoil for a Fixed-Wing UAV
Cruise speed, chord, endurance and how gently it stalls decide the section. A name that worked on a light aircraft is not automatically a UAV section.
- How to Design an Airfoil for Long-Endurance UAVs
Endurance is a high-lift efficiency problem at modest Reynolds number. The section has to hold that efficiency after the leading edge gets dirty.
- How to Generate Airfoil Coordinates and Export DAT, CSV and CAD Profiles
A section is a list of x and y. The file format, the order of the points, and the chord scaling decide whether the next tool sees the shape you designed.