Back to Projects
Sep 2024 – Jan 2025·Individual·Completed

Dolphin Kick Biomechanics

Mathematical Modelling · Biomechanics

Executed a biomechanical motion-tracking analysis to extract 72 discrete kinematic data points from underwater footage. Constructed a 39-equation piecewise model to map ankle trajectory and optimized the thrust-to-drag ratio via calculus to derive a theoretical maximum velocity of 2.64 m/s.

Mathematical ModellingBiomechanicsAdobe After EffectsKinematic Analysis

At a glance (one-pager)

Findings

Theoretical maximum velocity validated at 2.64 m/s

Optimization of the derived thrust-drag differential equation yielded a critical point at 2.64 m/s. This mathematically proves that the current kick frequency (b=1.7b = 1.7) is near optimal, isolating physical amplitude as the primary biomechanical bottleneck for performance gains.

Key Parameters

Tracking Duration

2.40s · 72 frames

GoPro at 29.97 fps; tracking ran until foot left frame.

Avg. Amplitude

1.50units ≈ 47.4 cm

Mean crest-to-trough across 4 peaks. Calibrated via foot length.

Avg. Frequency

1.7kicks/s

Fitted by Desmos slider; gives ~4 kicks over 2.40 s.

Measured Velocity

1.85m/s

Derived from x-displacement over ~30 frames (≈ 1 s).

Optimal Velocity

2.64m/s

From setting df/dv=0df/dv = 0 on the thrust–drag objective.

Piecewise Segments

39functions

Required to accurately model the full kick shape.

Objective

The underwater dolphin kick is heavily restricted in elite competition due to its unparalleled propulsive efficiency. The objective of this research was to extract empirical kinematic data from human athletic performance and construct a mathematical model precise enough to isolate the biomechanical variables governing optimal swimming velocity.

Methodology & reflection

Takeaways & position

Modeling human biomechanics exposed the severe limitations of idealized mathematical assumptions. The primary limiting factor of the optimization was assuming a direct proportionality between the amplitude-frequency product and actual velocity without an experimentally determined constant kk. Future kinematic modeling must incorporate Computational Fluid Dynamics (CFD) to measure true volumetric drag and utilize 3D motion tracking to account for lateral force vectors.

Methodology

Extracted 72 ankle coordinate pairs across a 2.40-second sample at 29.97 fps using Adobe After Effects motion tracking. Initial hypothesis testing proved a singular sinusoidal function insufficient due to multi-directional concavity reversals and near-vertical data intervals. Constructed a highly precise 39-equation piecewise continuous function using seven distinct algebraic families to map the physical geometry. This geometry was then generalized to extract mean amplitude and frequency parameters for calculus-based optimization.

Process

Formulated an objective function f(v)f(v) to balance forward thrust against quadratic fluid drag. Thrust was derived using mass flow rate principles yielding a linear term of 163v163v. Drag was modeled using literature-standard streamlined coefficients (CD=0.7C_D = 0.7) and measured cross-sectional area. Setting the derivative f(v)=16361.8vf'(v) = 163 - 61.8v equal to zero isolated the theoretical maximum velocity threshold.

Additional photos
Adobe After Effects motion tracking interface isolating 72 discrete coordinate pairs from 2.40 seconds of 29.97 fps underwater footage.
The 39-piece continuous algebraic model. Multi-directional concavity reversals proved that a standard sinusoidal approximation was mathematically insufficient.
Generalized sinusoidal approximation f(x) = 1.5sin(1.7x) + 0.147x + 2 superimposed on the raw data scatter plot to extract mean frequency and amplitude variables.

As per the University of Toronto Code of Academic Behaviour and the Professional Engineers Ontario Code of Ethics, I confirm that this project was completed individually and that the work described here is my own except where outside sources are cited.