科研速览 · Science Skim继续刷下去 · Keep skimming →
◆ Drying Technology2026-08-01· Reynolds-averaged Navier–Stokes equations

Multi-fidelity CFD surrogate framework for parametric analysis and multi-objective optimization of swirl flow in a co-current spray dryer air distributor

Tanakorn Rachapila

原始摘要(英文原文)· Original abstract
The swirl flow generated by the air distributor governs the chamber hydrodynamics of co-current spray dryers, influencing momentum distribution, residence-time characteristics, and flow stability. Earlier experimental and numerical work, most prominently from the University of Sydney group, identified a practical intermediate swirl setting rather than a universally steady operating band. Rather than redefining that range, the present study provides a computationally efficient, reliability-quantified design methodology for a specific small scale, food grade guide vane distributor, and positions its results against those earlier recommendations. A multi fidelity surrogate framework was developed that fuses 2D axisymmetric CFD (low fidelity, 50 Latin Hypercube samples) with targeted 3D periodic sector CFD (high-fidelity, 10 samples) through co-Kriging. Both fidelities employ steady-state RANS with the LRR Reynolds stress closure and a single phase gas model; consequently, the transient three dimensional swirl instability and the two phase evaporative load lie outside the present scope and are treated explicitly as limitations. Co-Kriging surrogates for Swirl Number, pressure drop, and velocity uniformity index achieved R2 > 0.99 for Swirl Number, a 34% RMSE reduction relative to single fidelity Kriging. Sobol analysis identified vane angle as the dominant factor for Swirl Number (first order index 0.81) and flow rate for pressure drop (0.63). NSGA-II optimization on the surrogate located a Pareto optimal window at α = 36–40° and Q = 1900-2,100 m³/h (S = 0.47–0.59, ΔP = 19–24 Pa), consistent with the practical swirl level of S ≈ 0.45 reported for co-current dryers, with ±0.015 prediction confidence on S. Pilot scale experiments with product feed (≈ 20% total solids) at the baseline geometry (α = 38°) validated the time mean thermal field and the integral mass and energy balances (velocity field MAPE ≈ 10.5%); the vane angle sweep itself is computational. The framework offers an accessible, open-source route to evidence-based distributor design for small to medium spray drying systems.
读原文 · Read the paper ↗

AI 追问PRO

登录后使用 AI 追问

讨论区

登录后参与讨论

相关论文 · Related

Multi-fidelity CFD surrogate framework for parametric analysis and multi-objective optimization of swirl flow in a co-current spray dryer air distributor — 科研速览 Science Skim