[Paper Review] Online Performance Optimization of a DC Motor Driving a Variable Pitch Propeller

Raphaël Cohen, David Miculescu+3 more|arXiv (Cornell University)|Sep 30, 2013

Advanced Combustion Engine Technologies4 citations

TL;DR

This paper proposes an online optimization algorithm that minimizes power consumption in a DC motor-driven variable pitch propeller system for a given thrust command. By iteratively adjusting the pitch angle using a variable step-size search, the system achieves a 26% reduction in power (from 10 W to 7.4 W) and reaches optimal performance in under 80 seconds, demonstrating practical feasibility for electric propulsion systems in UAVs and aerospace applications.

ABSTRACT

A practical online optimization scheme is developed for performance optimization of an electrical aircraft propulsion system. The goal is to minimize the power extraction of the propulsion system for any given thrust value. The online optimizer computes the optimum pitch angle of a variable pitch propeller by minimizing the power of the system for a command thrust value. This algorithm is tested on a DC motor driving a variable pitch propeller; the experimental hardware setup of the DC motor along with its variable pitch propeller is also described. Experimental results show the efficiency and practicality of the proposed online optimization scheme. Outstanding issues are sketched.

Motivation & Objective

• To develop an online, real-time optimization scheme that minimizes power extraction for a given thrust level in electric propulsion systems.
• To address performance degradation due to environmental changes, aging, and hardware variations in propulsion systems.
• To demonstrate the feasibility of online optimization using a DC motor and variable pitch propeller setup at zero airspeed.
• To improve convergence speed and optimality through a variable step-length pitch adjustment strategy.
• To lay the foundation for extending the method to complex aero engines such as turboprops, turbofans, and electric vertical takeoff systems.

Proposed method

• The system uses a DC motor with a variable pitch propeller, modeled via motor dynamics and blade element theory for thrust and torque.
• The optimization algorithm performs a gradient-free, iterative search over pitch angle to minimize input power for a commanded thrust.
• A fixed-step pitch search is first implemented, followed by a variable-step approach that accelerates convergence by adapting step size based on power change direction.
• The algorithm uses real-time measurements of power, thrust, and pitch angle to update the search direction and step size.
• The system is implemented on a physical testbed with feedback control to maintain commanded thrust during optimization.
• Simulations extend the results to non-zero airspeeds, validating the approach under varying flight conditions.

Experimental results

Research questions

• RQ1 Can online power optimization be effectively implemented in a DC motor-driven variable pitch propeller system under real-time constraints?
• RQ2 How does the convergence speed and optimality of the optimization algorithm depend on the choice of pitch step size?
• RQ3 To what extent can the system maintain commanded thrust while minimizing power consumption during the optimization process?
• RQ4 What are the performance gains of using variable step sizes compared to fixed step sizes in the optimization loop?
• RQ5 How does the system behave under non-zero airspeed and varying atmospheric conditions, and can the method be extended to full flight regimes?

Key findings

• The online optimization reduced power consumption from 10 W to 7.4 W—representing a 26% improvement—within the first 100 seconds of operation.
• The variable step-size approach achieved convergence in approximately 80 seconds, compared to 120 seconds with fixed steps, significantly improving response time.
• The system reached the commanded thrust of 0.52 N in about 15 seconds with variable steps, compared to 45 seconds with fixed steps, indicating faster tracking performance.
• The optimal pitch angle was found to be approximately 9 degrees, where the required power was minimized for the given thrust command.
• Power saturation occurred near 14 W, limiting the system’s ability to achieve higher thrust commands, indicating a hardware constraint.
• The experimental results confirm the practicality and efficiency of the online optimization scheme in real-world hardware, with consistent thrust tracking and power minimization.