Codes

The objective of this code is to exemplify how to improve the efficiency and accuracy of uncertainty analysis in aircraft conceptual design by using dynamic surrogate models combined with fuzzy clustering analysis to reduce computational cost while maintaining reliability. 

The code is based on Ref.: Du, S., & Wang, L. (2016). *Aircraft Design Optimization with Uncertainty Based on Fuzzy Clustering Analysis*. Journal of Aerospace Engineering, 29(1), 04015032.

This code uses the Breguet equation for an electric aircraft to estimate the sensitivity of the range to:

• empty mass fraction variation
• payload mass fraction variation;
• total mass variation;
• battery specific energy and technology level.

This code implements an integrated hesitant fuzzy MCDM approach for Quality Function Deployment (QFD), based on the method proposed by Wu, Liu & Wang (2016). 
It addresses key limitations of traditional QFD—namely, handling expert ambiguity, weighting customer requirements (CRs), and ranking engineering characteristics (ECs). 
The two‑stage algorithm combines hesitant fuzzy DEMATEL (to analyze interrelationships among CRs and compute their influential weights) with hesitant fuzzy VIKOR (to prioritize ECs via compromise ranking). 

Implementation steps include constructing collective hesitant fuzzy matrices, applying the OWA operator for defuzzification, building an influence relation map, and solving for ideal solutions. The model is validated on an electric vehicle product development case study. 
Outputs include CR weights, EC rankings, and a compromise solution, all under uncertain and conflicting criteria..

This notebook is designed to help you explore the effect of changing weights in the CST parametrization method. This method models smooth, flexible airfoil shapes by combining:

1. A class function that defines the general shape trend.
2. A Bernstein polynomial shape function scaled by coefficients.
3. An additive linear term for trailing edge thickness.

Ref.: Kulfan, B. M. (2007, January 8). A Universal Parametric Geometry Representation Method—"CST" (AIAA Paper 2007-62). 45th AIAA Aerospace Sciences Meeting and Exhibit, Reno, Nevada. https://doi.org/10.2514/6.2007-62

Wing optimization
This is a two-stage wing optimization for minimum drag of a wing.

Stage 1: Planform and twist optimization (using VLM, estimation of viscous drag, and IPOPT optimizer)

Stage 2: Airfoil shape optimization at the root, mid, and tip sections (Neuralfoil, and IPOPT)

The primary goal of incorporating uncertainty into the design process is to:
- Make performance insensitive to noises
-  Improve robustness in the manufacturing process
-  Ensure reliable system behavior under varying conditions
This approach helps in creating designs that are more resilient and reliable in real-world applications.

The code is based on the following Ref.:

1) Du, Shengchao, and Lifeng Wang. "Aircraft design optimization with uncertainty based on fuzzy clustering analysis." Journal of Aerospace Engineering 29, no. 1 (2016): 04015032.
2) Jaeger, L., Christian Gogu, Stéphane Segonds, and Christian Bes. "Aircraft multidisciplinary design optimization under both model and design variables uncertainty." Journal of Aircraft 50, no. 2 (2013): 528-538.

This code plots constraint diagram for a fixed-wing aircraft with conventional take-off and landing. 
The conditions considered are:

1. take-off;
2. turn;
3. climb;
4. cruise;
5. ceiling.

In addition, the code plots the wing loading for a range of maximum lift coefficients.

This code demonstrates:

• The types of measures of merit for various classes of aircraft
• How you might measure safety, reliability, maneuverability, or efficiency of an aircraft
• Approaches to compound merit function formulation

This code is full of beautiful radar charts!

The code employs five interconnected physical models to predict system performance:

1) Baseline aerodynamics model (I used Neuralfoil).

2) CFJ Duct Flow Model simulates the internal flow through the CFJ ductwork. It treats the duct as a 1D compressible pipe with area changes, computing pressure losses due to friction (via the Darcy‑Weisbach equation) and minor losses at bends, contractions, and expansions (using empirical loss coefficients). The model outputs the total pressure drop across the duct, which the compressor must overcome.

3)Compressor Model estimates the power required to drive the CFJ flow. It assumes isentropic compression with a fixed efficiency (75 %), calculates volumetric flow rate from mass flow, and uses a simplified quadratic curve to relate pressure ratio to mass flow rate. This allows the simulation to determine the actual power draw of the compressor given the total pressure rise needed.

4) CFJ Aerodynamic Enhancement Model predicts how the injected and suctioned flows modify the airfoil’s aerodynamics. Using empirical correlations, it computes the increase in lift and reduction in drag as functions of the momentum coefficient and angle of attack.

5) Momentum Coefficient Model serves as the coupling parameter between jet forcing and aerodynamic effect. It defines the momentum coefficient and inverts this definition to compute the required jet velocity and mass flow rate for a given coefficient. This links the compressor’s output to the desired level of flow control, closing the system loop.

This code implements an aircraft performance and optimization algorithm under uncertainty. 
It takes aircraft parameters (mass, geometry, aerodynamics, engine), uncertainty levels, technical specificationrequirements, and optimization settings as inputs. 
The algorithm performs an iterative equilibrium search to balance thrust and drag for speed determination, uses Monte Carlo simulation to account for random atmospheric conditions, and employs a search to guarantee global optimum in wing parameter optimization. 
Outputs include the probability of meeting TZ requirements for the baseline geometry, optimized wing parameters, performance improvement, atmospheric and performance plots, an optimization heatmap, and a detailed report with recommendations.

This code demonstrates the conceptual design of a day-night solar-powered aircraft.
It is built on simple mathematical models of the atmosphere, aerodynamics, solar irradiance, and flight mechanics. However, it employs iterative calculations of mass and energy balances to estimate the aircraft’s capability to perform multi-day missions, making it relatively robust. I enjoy exploring different design possibilities, so the code also includes a variety of trade studies. Enjoy :)

This code provides some foundation for evaluating winglet performance and conducting trade-off studies between aerodynamic benefits and structural penalties. 

This MATLAB code is designed to analyze the power requirements of a multi-motor aircraft, calculate the necessary battery specifications, and estimate the flight time based on the power consumption and battery capacity. The code takes inputs related to the aircraft's motors, thrust requirements, and battery specifications, and it generates graphical outputs to visualize the relationships between throttle, thrust, power, and current. 

This MATLAB code plots and analyzes using XFOIL one of the Selig's "bird-like" airfoils, AS6093. 
Ref.: Reference: Design of Bird-Like Airfoils by Gavin K. Ananda∗ and Michael S. Selig

This MATLAB function computes key parameters of the Atmospheric Boundary Layer (ABL) for wind flow simulations, including:
- Mean wind speed profile (power law)
- Turbulence intensity profile (piecewise-defined)
- Turbulent kinetic energy (TKE)
- Turbulence dissipation rate
These parameters are important for computational fluid dynamics (CFD) simulations of wind flow around structures (e.g., buildings, wind turbines).

This MATLAB code is designed to compute and visualize the pressure coefficient (( Cp )) distribution across an airfoil using data from XFoil or a similar aerodynamic analysis tool. The code takes various input parameters related to the airfoil and flow conditions, processes the data, and generates a graphical representation of the ( Cp ) distribution. 

One of the first tasks in any new aircraft design is to perform a constraint analysis using a special graph called a constraint analysis graph. The primary advantage of this graph is that it can be used to assess the required wing area and power plant for the design, such that it will meet all performance requirements. Constraint analysis is used to assess the relative significance of performance constraints on the design. This is done by plotting the constraints on a special two-dimensional graph called the design space. Commonly the two axes represent characteristics such as (y-axis) thrust-to-weight ratio (T/W) and (x-axis) wing loading (W/S). The graph is then read by noting that any combinations of W/S and T/W that are above the constraint curves will result in a design that meets those requirements. 

This script visualizes and analyzes the relationship between mass and wing loading across different categories of flying objects, such as birds, UAVs, solar-powered aircraft, and insects. 
Ref.: Noth, A. (2008). “Design of Solar Powered Airplanes for Continuous Flight.”. Url: [PDF] Design of Solar Powered Airplanes for Continuous Flight.

This is a collection of scripts intended to help you optimize an airfoil for the prescribed limitations and measures of merit. This program calls optfoil.exe to perform optimization, which in turn utilizes the Particle swarm and Simplex optimization algorithms, as well as the xfoil.exe for the aerodynamic analysis. 

The script is intended to:
- provide the basic features of the generalized Hicks-Henne function;
- show the effects of the input paramters on the outpit of the generalized Hicks-Henne function.

This MATLAB script automates the selection of a motor, propeller, and battery pack for a model aircraft based on user-defined parameters such as weight, desired speed, and thrust requirements.  

In the archive, you will find:
1) a code Run_prop_mass_prediction_model.ipynb  predicting propeller weight (in grams) based on diameter, material, and  blade count using two machine learning models trained on 114 propellers from  APC, Mejzlik, Xoar, and T-Motor;
2) both ML models (combined_general_5_60.pkl and combined_carbon_10_40.pkl);
3) dataset with 114 propeller data (from various sources on the Internet);
4) a README.txt file with the usage guidelines. and ML model description.

INPUT PARAMETERS
- Diameter (inches): 5-60 inches recommended range
- Material: Composite, Carbon, Beechwood, or Aluminium
- Blades: 2, 3, or 4

OUTPUT
- Estimated weight in grams (1 decimal place)
- Model used for prediction
- Expected accuracy range
- Warnings if outside training range

MODELS USED
- General Model: Gradient Boosting (5-60" diameter, 15% typical error)
- Carbon Model: Random Forest (10-40" diameter, 6-8% typical error)
- Fallback: Power law formulas (all sizes, 20-25% typical error)

ACCURACY NOTES
- Best accuracy: Carbon props 10-40 inches (6-8% error)
- Good accuracy: All props 5-60 inches (15% error)
- Acceptable accuracy: Outside ranges using formulas (20-25% error)

DEPENDENCIES
- Python 3.8+
- numpy, pandas, scikit-learn, joblib

FILES REQUIRED
- Run_prop_mass_prediction_model.ipynb (main script)
- combined_general_5_60.pkl (general ML model)
- combined_carbon_10_40.pkl (carbon ML model)

METHODOLOGY
The models were trained on  propeller data using:
- Feature engineering (diameter^2, diameter^3, log transformations)
- Gradient Boosting for general propellers
- Random Forest for carbon propellers
- Cross-validation to prevent overfitting
- Material factors derived from density ratios