Datasets for "Evolutionary Steering of Fluidized Bed Hydrodynamics via Spatial Modulation of Inlet Gas Velocities"

This repository contains numerical datasets generated for the evolutionary steering of inlet gas velocities in quasi-two-dimensional fluidized beds (FBs). All simulations apply a sinusoidal, time-dependent perturbation to the mean inlet velocity. The oscillatory inlet profiles are fully generated by evolutionary algorithms (EAs), requiring only the mean operating velocity as input. The dataset includes single-objective (bubble-size control) and multi-objective (simultaneous bed-expansion and bubble-diameter control) optimization studies. All cases were performed using a static bed height of 0.10 m, an operating gas flow rate of 16 L min⁻¹, and video acquisition at 50 fps. EA settings: Single-objective hyperparameter study: population size 10, 10 generations Multi-objective optimization (NSGA-II, R-NSGA-II): population size 10, 20 generations Random-walk baseline: 1 generation, 200 candidates Repository Structure ```Main_Repository/├── Hyperparameter_Study/│ ├── Final_Screening/│ │ ├── Bubble_Dynamics/│ │ │ └── cxpb{value}_mutpb{value}/│ │ │ └── run{value}.csv│ │ └── EA_Stats/│ │ └── cxpb{value}_mutpb{value}/│ │ └── run{value}.csv│ └── Preliminary_Screening/│ ├── Bubble_Dynamics/│ │ └── cxpb{value}_mutpb{value}/│ │ └── run{value}.csv│ └── EA_Stats/│ └── cxpb{value}_mutpb{value}/│ └── run{value}.csv└── Multi-objective_Steering/ └── Algorithm: NSGA-II, R-NSGA-II, Random walk/ ├── Bubble_Dynamics/ │ └── run{value}.csv └── EA_Stats/ └── run{value}.csv``` Bubble_Dynamics/ Stores the processed bubble-tracking results for each oscillatory inlet profile.Each file is named run{value}.csv, representing an independent EA evaluation. Fields: video_name – User-defined video identifier gen – EA generation frame_number – Frame index time [s] – Time stamp num_contours_per_frame – Number of detected bubbles ID – Unique bubble ID cx_pos [px], cy_pos [px] – Bubble centroid (pixels) cx_pos [m], cy_pos [m] – Bubble centroid (SI units) size [px^2], size [m^2] – Bubble area perimeter [px], perimeter [m] – Bubble perimeter approx_diameter [px], approx_diameter [m] – Estimated bubble diameter x_velocity [m/s], y_velocity [m/s] – Bubble velocities between frames valve1 … valve5 – Inlet velocities per valve (L/min) EA_Stats/ Contains the complete EA history for each individual of each generation, using the same run{value}.csv filenames to allow direct correlation with Bubble_Dynamics. Fields: gen – EA generation v1 … v5 – Valve-specific inlet velocities (L/min) F, F1, F2 – Fitness values F for single-objective F1 (bed-expansion deviation) and F2 (bubble-size deviation) for multi-objective CV – Total constraint violation H – Equality constraint G1, G2 – Inequality constraints Hyperparameter Study Two subdirectories explore crossover and mutation probabilities: Preliminary_Screening:7 independent runs for each pair(pc,pm)(p_c, p_m)(pc,pm), wherepc∈{0.1,0.2,…,1.0}p_c \in \{0.1, 0.2, \dots, 1.0\}pc∈{0.1,0.2,…,1.0} andpm∈{0.0,0.001,0.01,0.05,0.1}p_m \in \{0.0, 0.001, 0.01, 0.05, 0.1\}pm∈{0.0,0.001,0.01,0.05,0.1} Final_Screening:Selected (pc,pm)(p_c, p_m)(pc,pm) pairs with 31 independent runs each Both include full bubble dynamics and EA histories. Multi-objective Steering Contains the same data structure as above but for three algorithms: NSGA-II R-NSGA-II Random-walk baseline Funding This work is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), SPP 2364, project number 504548777.