TL;DR
This paper introduces a novel agent-based model for fish schooling that explicitly incorporates the hydrodynamic wakes generated by swimming fish. By integrating these complex fluid dynamics with social behaviors, the model reveals how these wakes significantly influence school organization, leading to more structured patterns, especially when social forces are weaker.
What Problem Does This Paper Address?
Fish schooling is a fascinating collective behavior with clear benefits for individual fish, such as enhanced foraging and predator evasion. While existing models have successfully captured emergent schooling patterns using rules for attraction, alignment, and avoidance, they often neglect the crucial role of hydrodynamics. Specifically, the complex vortex wakes shed by swimming fish, which can extend for multiple body lengths, carry vital information and exert forces on neighboring fish. Previous models that only considered potential flow effects failed to capture these persistent vortex wakes. This study aims to bridge this gap by developing a more realistic hydrodynamic model for fish schooling to understand how these fluid interactions shape collective behavior.
Methods
The authors developed a novel agent-based model that couples a Newtonian physics-based framework for fish movement with an ansatz for sensing and social interactions. The model simplifies fish to 2D entities swimming in a horizontal plane but uses 3D direct numerical simulations (DNS) to parameterize the hydrodynamic forces and torques, including both potential flow and vortex wakes.
Key methodological components include:
- Force-Momentum Balance: Fish motion in surge, sway, and yaw is governed by Newton's second law, incorporating propulsive forces, hydrodynamic forces/torques, and social interaction forces/torques.
- Hydrodynamic Wake Model: This is the core novelty. It includes:
- Potential Flow: Modeled using an ellipsoid approximation with sink-and-source pairs.
- Vortex Wake: Synthesized from DNS data, representing the complex, counter-rotating vortex pairs shed by the caudal fin. This is a significant advancement over simpler potential flow models.
- Social Interaction Model: Incorporates attraction, alignment, and repulsion forces and torques, controlled by parameters that can be tuned to represent different behavioral tendencies (e.g., high-alignment vs. high-attraction).
- Visual Sensing: A cardioid-shaped vision field with "attention parsimony" is used, meaning fish only track a limited number of neighbors.
- Numerical Solution: Equations are solved using a 4th-order Runge-Kutta scheme in MATLAB.
- Analysis: Principal Component Analysis (PCA) is used to analyze the emergent schooling topologies and quantify the impact of wake dynamics.
Datasets
The paper relies on data generated from:
- 3D Direct Numerical Simulations (DNS): Used to parameterize the hydrodynamic forces and moments (drag, lift, yaw torque) and to characterize the vortex wake structure of a carangiform swimmer.
- Ensemble Simulations: The agent-based model itself is used to generate data by running 300 simulations for each scenario to obtain statistically reliable results.
No external datasets for training or evaluation were explicitly mentioned, as the model is built upon simulated data and parameterized by DNS.
Key Insights and Ideas
- Wakes are Crucial for Organization: The most striking insight is that hydrodynamic wakes, not just social forces, are critical for organizing fish schools, especially when social alignment is dominant.
- "Diamond" Pattern Emergence: In high-alignment scenarios, the wake creates favorable drifting regions that passively attract trailing fish, leading to a distinct "diamond" pattern. This suggests that hydrodynamic efficiency can drive emergent structures.
- Wake Strength Matters: There's an optimal range for vortex strength. Too weak, and the hydrodynamic influence is minimal; too strong, and it can destabilize the school.
- High-Attraction vs. High-Alignment: The impact of wakes differs significantly. In high-alignment, wakes create structured patterns. In high-attraction, individual fish move more dynamically, making it harder for wakes to superpose and create consistent structures.
- 2D Simplification is Valid: The study demonstrates that a 2D model, when incorporating realistic hydrodynamic wakes, can capture essential schooling dynamics, supporting its use as a first-order approximation.
Key Results
- Improved Schooling Topology: The inclusion of vortex wakes leads to more ordered and structured schooling patterns, particularly in high-alignment conditions.
- Quantitative Impact of Wakes: PCA analysis showed that the first principal component (representing the dominant pattern) explained a significantly larger portion of the variance in schools with wakes compared to those without (e.g., 11.95% vs. 1.54% for high-alignment).
- Wake Width Influence: Wider wakes (lower Re or AR) enhance schooling organization in highly polarized groups (high-alignment), while narrower wakes show less distinction between high-alignment and high-attraction.
- Wake Strength Optimization: Increased vortex strength generally leads to more organized patterns in high-alignment modes, up to a certain point, after which it can cause destabilization.
Limitations and Future Work
- 2D Simplification: While effective, the model is restricted to two dimensions. Future work could extend it to three dimensions for a more comprehensive understanding of complex schooling structures.
- Sensory Modalities: The model currently only includes vision-based sensing. Incorporating other senses like pressure and flow sensing could enhance the realism of simulated fish interactions.
- Species-Specific Parameterization: The model is flexible and can be parameterized for different species, but further work could explore a wider range of environmental conditions and species-specific behaviors.
- Experimental Validation: While the model is informed by DNS, direct experimental validation of the predicted schooling patterns would be valuable.
Summary updated