Propeller Perfection: Optimization Using JMP for Efficient Experimentation

brycerjs8 · Feb 18, 2025 10:12 AM

Efficient propeller design is crucial for the electric Vertical Takeoff and Landing (eVTOL) industry, which is projected to reach a market value of $15.8 billion by 2030, according to a report by Market Research Future. Propellers play a key role in optimizing performance, reducing energy consumption, and enhancing flight stability in eVTOL aircraft, which are designed to revolutionize urban air mobility. In order to achieve these goals, using a space-filling design (a type of design of experiments, or DOE) helps explore the complex relationship between design variables, ensuring the most efficient configurations. For this project, Blade Element Theory was employed to model the aerodynamics of the propeller, while integration with MATLAB and JMP software provided a powerful platform for conducting the analysis and optimizing the design. Through this approach, the impact of different variables on propeller performance can be better understood and refined for the growing eVTOL industry.

Byron_JMP · ‎02-20-2025

Nice presentation!

what would be really cool is if you could take the parameters from the Profiler and feed them directly into a parametric model in Fusion 360 so that models could be physically validated.

statman · ‎02-20-2025

I did not get a chance to review all of the attached files, so you might want to ignore my comments.

Correct me if I'm wrong, the "experiments" were run in simulation software? I don't understand why you do this? The model is already known in the simulation software. Are you trying to uncover that model? How was error introduced? How was that model developed (e.g., completely theoretical or the result of experimentation)? What inference space was the model developed in? How was noise simulated?

What Byron suggests is much more important IMHO. Physically validate the model in the simulation software.

brycerjs8

@statman Thank you for watching and commenting!

The model used is called Blade Element Theory There are many ways to implement this model, for this experiment we introduced variables that are typically assumed to be linear in Blade Element Theory, meaning the model is not already known.

The model was developed in 1878 and has been validated experimentally, it is the most commonly used analytical model to simulate propeller performance.

We limited the experiment to 6 input variables and put constraints on each of them based on theoretical knowledge and physical limitations.

For example, a tip pitch angle greater than 17 degrees would cause all 5 airfoils to be in their stall region.

Since this experiment is purely analytical, the results will be the same no matter how many times you run it, meaning there is no need to simulate noise or introduce error.

Further experimentation would include physical validation or high-fidelity simulation tools, the purpose of this project is to demonstrate JMPs ability to uncover results that would otherwise be unexplored and to demonstrate a use case of integrating JMP and MATLAB.

statman

Thanks for your reply. I still don't understand, so please help me understand. You are not actually doing any physical experiments, correct? You must be using theoretical models in your software? When you input a value for a variable, if that variable is not in the theoretical model, then it will have no effect on the response variables, correct? In order for there to be a change in the response in the simulation, the variable must nbc in the model. If there is only a linear term in the simulation software, the software can't possibly show a non-linear effect, correct. The model is known.

Having spent years working in aerospace and, unfortunately on the V-22 (Osprey), I can assure you the importance of noise in understanding how to "perfect" propellor performance. It is impossible to manufacture 2 blades the same (these are carbon fiber autoclaved blades). The assumptions of your software are your inputs are exactly as input without variation....this is not possible in reality.

I understand, perhaps, you don't want to understand really what factors affect the performance of a rotating blade, but instead want to show the power of using JMP and Matlab.

BTW, do you have any physical evidence that when all blades are >17º there is stall? In your attached link "The critical angle of attack is typically about 15°, but it may vary significantly depending on the fluid, foil – including its shape, size, and finish – and Reynolds number."