Regression Analysis of Aerodynamic Characteristics for 87 NACA 5-Digit Airfoil Classes

Regression analysis was employed to investigate the complex relationship between the aerodynamic characteristics of NACA 5-digit airfoils and key predictors, including angle of attack (alpha), Reynolds number, optimal lift coefficient (X1), maximum camber (X2), airfoil thickness at the maximum camber (S or X3), and maximum thickness position (Max Thickness or X4). The objective was to develop robust predictive models for critical aerodynamic responses, namely lift coefficient (CL), drag coefficient (CD), profile drag coefficient (CDp), minimum pressure coefficient (Cpmin), and pitching moment coefficient (Cm).

By leveraging regression analysis, we aimed to uncover the nuanced dependencies and intricate patterns within the data, allowing for the creation of empirical models that capture the influence of the selected predictors on the aerodynamic performance of the airfoils. Understanding these relationships is crucial for optimizing airfoil design, enhancing aerodynamic efficiency, and gaining insights into the complex interplay of factors affecting the lift, drag, pressure distribution, and pitching moment characteristics of the airfoils across a wide range of operating conditions. The resulting regression models provide a valuable tool for predicting and optimizing the aerodynamic behavior of NACA 5-digit airfoils, contributing to advancements in aerodynamic design and performance analysis in various applications, such as aviation and wind energy. 

In summary, the table provides a comparative overview of the performance of different regression models in predicting Inputs vs CDp Response. Models are assessed based on various metrics, with lower RMSE, MSE, and MAE values and higher R-squared values suggests that these models have been evaluated on a separate test dataset indicating better predictive performance. The "Tested" status metrics, with lower RMSE, MSE, and MAE values and higher R-squared values indicating better predictive performance. The "Tested" status suggests that these models have been evaluated on a separate test dataset.

Author: Andrew Clinton

PCA Insights: Analyzing 87 Classes of NACA 5-Digit Airfoils in the Project

 

When considering all six variables, the Correlation plot that is derived appears to be inconsistent, with some variables showing minimum to no correlation. After observing this behavior, we can conclude that some variables are negligible and we extract a Pareto Chart from MATLAB to deduce how many principal components should be considered in the analysis.

The Pareto chart informs us that the first four variables are sufficient in achieving close to 100% variance and therefore we neglect two of the variables and shift our attention to the four principal components for Principal Component Analysis (PCA).

A biplot is a type of plot that combines information from both the observations (rows) and the variables (columns) in a multivariate dataset. It is particularly useful for visualizing relationships and patterns in high-dimensional data. The biplot displays points for each observation and vectors for each variable. The position of an observation point relative to a variable vector provides insights into the relationships between variables and observations. The length and direction of the vectors indicate the strength and direction of the variable's influence.

Once Principal Component Analysis is conducted and the four principal components are analysed, the correlation plot takes a more convincing structure and shows consistent correlation between the variables.

Author: Shaad Akbar & Andrew Clinton

Decoding Aerodynamics: Unveiling the Secrets of NACA 5-Digit Airfoils through Regression Analysis

Welcome to this short video where we'll explore how regression analysis was employed to understand the aerodynamic characteristics of NACA 5-digit airfoils. In our study, we focused on 87 airfoil classes, using key predictors such as angle of attack, Reynolds number, optimal lift coefficient, maximum camber, airfoil thickness at the maximum camber, and maximum thickness position.

Our goal was to unravel the intricate relationships between these predictors and critical aerodynamic responses: lift coefficient (CL), drag coefficient (CD), profile drag coefficient (CDp), minimum pressure coefficient (Cpmin), and pitching moment coefficient (Cm). By leveraging regression analysis, we aimed to create predictive models that capture the complex interplay of factors influencing airfoil performance.

At the end of the video, you'll discover the practical side of our analysis – the deployment of these models.

Author: Andrew Clinton

PCA Unraveled: A Humorous Dive into Data Analysis

 A short video on PCA. Enjoy😁

Author: Andrew Clinton

Data extraction

We both were given 100 airfoils to analyze using XFOIL/XFLR5. We split the airfoils to 50 each and proceeded to extract data using these software. Inputs were fed into the XFLR5 software whereby the NACA geometries, the angles of attack and the Reynolds Numbers were entered. The outputs were manually saved and stored into a databank of more 1500 files. These will then be batch processed using MATLAB this week. Andrew will update a video of this process.

Overview of Project

 Airfoils are cross sectional geometries of plane wings that are able to induce lift upon the body of an aircraft due to their shapes and orientations. Airfoils have been developed extensively over the last decade or so and can be found in several different categories. For the purpose of the MM466 Project, the team will be working on the NACA 5-Digit airfoil geometries to obtain data sets.

Author: Andrew Clinton