Desiccation, Crust Formation & Consolidation of Soft, Cohesive Soil due to Atmospheric Conditions

Abstract

Deterioration of the ecological quality within the Markermeer, due to the elimination of natural flushing and a resulting increase in turbidity, has created a need to revitalize the region for the benefit of the local wildlife. One major way in which this is being achieved is through the Marker Wadden, designed as a wetland area featuring a combination of islands, marsh-pond areas, mud flats, and sheltered shores which will serve as bird and wildlife spawning area. One of the primary innovations within the Marker Wadden project is the utilization of locally dredged material to form many of the wetland areas. This dredged material −often referred to as mud− is characterized as a soft, cohesive, fine-grained soil material with a mixed sediment composition of clay, silt, fine sand, organic material, water, and often gas (Winterwerp and van Kesteren, 2004). This material has a lot of inherent uncertainties in regards to its behavior and use in construction, stemming from its general non-homogeneity and less predictable/studied behavior. When considering the use of this material to construct areas with specifically designated elevation and strength ranges, the additional strength and settlement effects due to desiccation and crust formation once the material surface has been exposed to atmospheric conditions must be carefully considered. The assessment of this process includes the outlining of a conceptual model emphasizing two distinct stages of desiccation, lab testing to define material behavior during this phase, large-scale testing, and finally an evaluation of an existing numerical model for this process. The conceptual model was then compared and validated in part through both physical, in-situ measurements from a large-scale test setup, as well as to a numerical model developed at Delft University of Technology and further adapted for this process. In addition, the numerical model was utilized as a prediction tool for the Marker Wadden, under various atmospheric conditions and surface water conditions. These effects were evaluated in terms of water content, void ratio, settlement, and strength.

The large-scale test setups −representing three varying degrees of atmospheric conditions and surface water conditions− exhibited similar trends in both the in-situ measurements and modelled results, with increasing levels of active surface water removal and precipitation minimization correlating to increased void ratio profile reduction −and therefore increased desiccation and drying− especially at the surface. However, due to the short-term nature of the large-scale testing, none of the large-scale tests ever reached the second stage of desiccation, and it can be concluded that in the initial stage of desiccation, minimal shear strength development will occur in the surface material. Modelling of variations in surface water conditions show that active removal of surface water promotes faster progression into the second stage of drying, and therefore faster initial development of shear strength. Other significant findings include the long-term stabilization in terms of total additional settlement, and that neither short-term re-submergence of the material surface nor initial seasonal start date will hinder this long-term settlement stabilization. Furthermore, the reality of the large-scale test material, including the presence of increased sand content layers, also highlights the importance of assessing the profile material variation, as deviation between the various in-situ measurement profiles and modelled “uniform” profile results show a high deviation. Modelling of this process is limited to the selection of the proper material input values utilized for the complete profile. During this modelling process, it was determined that the special consideration should be given to how the soil water retention properties are obtained, as well as the hydraulic conductivity relation and values utilized, as small deviations in these inputs impact the model significantly in comparison to other inputs.