Volume 31, Issue 10, Supplement, 15 July 2011, Pages S84–S93

Proceedings of the 9th International Conference on Nearshore and Estuarine Cohesive Sediment Transport Processes

Edited By Pierre Le Hir, Erik Toorman, Florence Cayocca and Romaric Verney

Modeling flocculation processes: Intercomparison of a size class-based model and a distribution-based model

  • a Institute for Chemistry and Biology of the Marine Environment, University of Oldenburg, POB 2503, 26111 Oldenburg, Germany
  • b Ifremer, Hydrodynamics and Sediment Dynamics Laboratory (DYNECO/PHYSED), BP 70, 29280 Plouzané, France
  • c Institute for Coastal Research, GKSS Research Center, Geesthacht, Germany

Abstract

Modeling suspended particulate matter (SPM) dynamics is essential to calculate sediment transport budgets and to provide relevant knowledge for the understanding of biogeochemical cycles in coastal waters. Natural flocs are characterized by their size, shape, structure and density that determine their settling velocity and therefore their vertical as well as horizontal transport. During transport, several processes, in particular aggregation and fragmentation, alter these particle properties. In the present study, we compare two different 0D modeling approaches for flocculation processes, a size class-based (SCB) model and a distribution-based (DB) model that follows the first moment of the particle distribution function. The study leads to an improved understanding of both models, which aim to better resolve SPM dynamics in spatial and ecosystem models in the near future. Both models are validated using data from laboratory experiments. The time evolution of the particle dynamics subjected to tidal forcing is represented equally well by both models, in particular in terms of (i) the mean diameter, (ii) the computed mean settling velocity and (iii) the particle size distribution. A sensitivity study revealed low sensitivity to changes in the collision efficiency and initial conditions, but a high sensitivity with respect to the particles’ fractal dimension. The latter is an incitation to enhance the knowledge on processes related to changes of fractal dimension in order to further improve SPM transport models. The limitations of both models are discussed. The model intercomparison revealed that the SCB model is useful for studies focussing on the time evolution of floc distributions, especially under highly variable conditions. By contrast, the DB model is more suitable for studies dealing with larger spatial scales and, moreover, with coupled marine physical–biogeochemical systems, as it is computationally very effective.

Keywords

  • Suspended particulate matter (SPM);
  • Flocculation/aggregation;
  • Size class-based model;
  • Distribution-based model;
  • Tidal dynamics;
  • Model comparison

Figures and tables from this article:

Full-size image (19 K)
Fig. 1. 

Aggregation of flocs of mass mi and mj into mass mi+mj and subsequent distribution into the nearest size classes by using a mass-weighted interpolation. Vi and ρi are the volume and the effective density of a floc in size class i, respectively.

Full-size image (24 K)
Fig. 2. 

Mean floc size variation during a simulated tidal cycle: comparison between laboratory measurements and models results. Note that the mean diameter of the observed flocs might be overestimated due to limitations of the camera system. SCB: size class-based and DB: distribution-based models.

Full-size image (53 K)
Fig. 3. 

Normalized concentration in % of the experimentally derived distribution (EXP), size class-based (SCB) model and distribution-based (DB) model. Note that both models fail for the time View the MathML source as settling is not taken into account in the model comparison.

Full-size image (37 K)
Fig. 4. 

Normalized concentration in % of the experimentally derived distribution (EXP), size class-based (SCB) and distribution-based (DB) models for different times (t).

Full-size image (24 K)
Fig. 5. 

Mean settling velocity for experimentally observed particles, size class-based (SCB) model and distribution-based (DB) models calculated by Eqs. (21) (data, SCB model) and (22) (DB model). Note that both models fail for the time View the MathML source as the models do not account for changes in the size distribution by settling.

Full-size image (29 K)
Fig. 6. 

Left: time evolution of the mean floc diameter as simulated by the distribution-based (DB) and the size-class based (SCB) models. Right: floc size distribution computed by both models at simulation start and after 20 min (indicated by open circles in the left panel). Notice that only the term of aggregation is taken into account with the same initial distribution and the same parameter set used in the comparison with data (see Table 1).

Full-size image (19 K)
Fig. 7. 

Sensitivity of both models to initial conditions. SCB: size class-based model; DB: distribution-based model.

Full-size image (21 K)
Fig. 8. 

Sensitivity of both models to changed fractal dimension compared to the reference runs. SCB: size class-based model; DB: distribution-based model.

Full-size image (19 K)
Fig. 9. 

Sensitivity of both models to variations in collision efficiency compared to the reference runs. SCB: size class-based model; DB: distribution-based model.

Full-size image (33 K)
Fig. 10. 

Growth rates for the mean radius of the DB model. (A) Different α and fB pairs for the same steady state (View the MathML source). The correspondent α:fB values are: 0.04:2801 s1/2 m−2, 0.11:7701 s0.5 m−2, 0.18:12602 s0.5 m−2, 0.25:17503 s0.5 m−2, and 0.32:22404 s0.5 m−2. (B) The same α and fB pair as for the reference run, but using different shear values.

Table 1.

Parameter set for the reference runs.

In case of the break-up factor fB and the collision efficiency α values are given as DB/SCB values.

View Within Article
Corresponding author contact information
Corresponding author. Tel.: +49 441 798 3634; fax: +49 441 798 3404.