Abstract
This paper discusses the transport of suspended solids (SS) by horizontal large eddies (HLE) from a narrow channel to surrounding vegetation areas in a constructed wetland. A series of flume experiments were carried out for several conditions of width/depth ratio and Reynolds number, referring to the ranges of parameters reported in actual constructed wetlands. The unsteady velocity field was measured in detail using the Particle Image Velocimetry (PIV) technique. Numerical simulations were conducted under the same conditions as the experiments, and computation results agreed with observation results very well. The study clarified the dependency of HLE scale and intensity on the aforementioned parameters. Numerical tracer experiments were conducted to investigate the basic characteristics of SS transport by HLE. The results show that the range of SS intrusion to vegetation areas is comparable with channel width. This fact suggests that a system of channels should be designed to distribute SS in a wide area of vegetation in a constructed wetland.
Introduction
Constructed wetlands are vital for treating urban runoff and wastewater, primarily by allowing suspended solids to settle. Ideally, water should move uniformly through the wetland to maximize residence time. However, "short-circuiting" often occurs, where water carves out preferential flow paths (channels) through the vegetation. These paths can carry up to 70% of the water through the system in a fraction of the intended time, severely degrading treatment efficiency. This study investigates whether intentionally designed artificial channels can create specific turbulence patterns—horizontal large eddies (HLE)—at the boundary between open water and plants to distribute pollutants more effectively across the entire wetland area.
Methodology
The researchers employed a dual-track approach combining physical flume experiments with numerical modeling. A 10-meter-long recirculating flume was used, with iron nails arranged in a staggered pattern on both sides of a central open channel to simulate wetland vegetation.
Surface velocity fields were mapped at 30Hz using sawdust tracers and PIV techniques. Simultaneously, a quasi-3D flow model accounting for vegetation resistance was used for numerical simulations. To simulate pollutant transport, a "Random Walk" model tracked 10,000 virtual particles as they interacted with the eddies. The experimental dataset included nine runs (A-1 through C-3) varying flow rates (26 to 56 l/min), water depths (2.6 to 3.8 cm), and channel widths (12 cm or 20 cm), with Reynolds numbers ranging from 1,000 to 2,400 to match real-world conditions.
Results and Discussion
A key finding was that the wavelength of horizontal eddies is primarily dictated by the width of the channel rather than the flow velocity. Turbulence is strongest at the interface where fast-moving channel water meets slow-moving vegetated areas; this velocity gradient creates shear-unstable horizontal eddies that physically transport water and pollutants laterally.
The numerical simulations showed excellent agreement with the physical flume experiments regarding eddy scale and intensity. However, numerical analysis revealed that the lateral intrusion of pollutants into the vegetation is limited, roughly comparable to the width of the channel itself. In simulations, while 70% to 80% of tracers eventually entered the vegetation zones, they did not travel deep into them. The "90% intrusion distance" ($L_{90}$) remained remarkably consistent at around 15 cm across different cases, which is close to the channel widths used.
Conclusion
Because lateral transport distance is limited by channel width, a single straight artificial channel is insufficient for maximizing treatment area. To utilize the full area of a wetland, channel systems must be designed with branching or meandering paths to ensure eddies reach all corners of the vegetation. Future work should address the behavior of real suspended solids, which may settle faster than neutral tracers once pushed into low-velocity vegetation zones, potentially increasing system efficiency beyond what tracer models suggest.
Summary updated