scope of the study


The main objective of the study was to develop and apply laboratory instrumentation that enabled the direct measurement of the semipermeability of natural contaminated materials. The semipermeability of natural clays, its importance and possible impact on the transport of contaminants, was first discussed in the Netherlands within the framework of the environmental aspects of permanent harbour sludge depots. One of the main depots, De Slufter near the Port of Rotterdam, is constructed in a salt water aquifer without a bottom liner. The clayey harbour sludge itself is thought to act as a low permeable layer, restricting advective transport of contaminants. However, the design possibly also enables the sludge layer to act as a semipermeable membrane. Semipermeability can induce water transport by chemical osmosis, and subsequently transport of contaminants out of the sludge into the aquifer. As a consequence, the design criteria regarding water and pollutant fluxes from the depot as laid down in Dutch legislation may be exceeded. Other man made situations where semipermeability of clays can have a profound effect on the transport of contaminants may exist in landfill sites. Here clay barriers, often consisting of bentonite or a sand–bentonite mixture, are applied to prevent emissions from the landfill. Against this background the samples investigated in this study are natural clays —bentonite— and clayey contaminated sediments —harbour sludge— under conditions closely related to those encountered in the Netherlands.

On this page you will find a condensed version of the synthesis as included in the thesis. In this synthesis all topics addressed during the research are briefly described and the findings of this research presented.



performance of the experimental design


The major advantage of our experimental design in comparison to those used in most earlier studies is the direct measurement of chemical osmosis in a sample. This enables the study of the evolution of semipermeability over time, as was done in one of the experiments on the harbour sludge. This clearly shows that σ is not an intrinsic property of a given sample at given conditions. The semipermeable properties change as a result of other processes such as diffusion and cation exchange that proceed simultaneously with osmosis whenever the sample is a non–ideal membrane. These processes also occur in the subsurface when a clay or a clayey sediment is subjected to a chemical gradient. The experimental design therefore allows a laboratory study of coupled flow phenomena in a setting more closely related to the Dutch field situation. The use of a flexible wall permeameter (Figure 1) allows the simulation of various overburden loads by adjusting the pressure in the cell liquid.

Problems elsewhere encountered with direct measurement of chemical osmosis like insufficient mixing of the reservoirs as described in the literature were overcome. The flexible permeameter design, however, has its own limitations which should be overcome in future studies. One of these problems is leakage from the pressurised water in the cell to the reservoirs.

Of the criteria formulated for a experimental design to measure chemical osmosis all but one were met. The choice of stainless steel components did not prove to be satisfactory. Although the reservoirs were free of leakage and more rigid than expected corrosion of the stainless steel reservoirs did occur. Cleaning the reservoirs between experiments proved to be tedious and time consuming but necessary to eliminate the possibility of erroneous electrical conductivity signals.

Photo 3_2

Figure 1
Photo showing the flexible wall permeameter (Keijzer, 2000).


In our study the length of an experiment was not a major requirement. Generally, it took three months to complete an experiment on a given sample, including preparation of the experimental setup between samples, calibration of the pressure transducers and electrical conductivity cell, and determination of the hydraulic conductivity. Although, not considered to be a major drawback, this type of design is not suited for a quick screening on semipermeability of natural clayey materials.



implications for the storage of contaminated sludge


Because harbour sludge can act as a semipermeable membrane and hydrophobic contaminants do desorb under low convective flow regimes in these sediments, the storage of conntaminated sediments in sludge depots does not eliminate the risk of further dispersion in the environment.

Currently, the sludge layer in De Slufter approaches a thickness of 20 m. The water expelled from the sludge is present as a fresh/brackish layer on top, and a lens in the salt water aquifer underneath the depot. Over time this lens will dissipate by diffusion and horizontal convective transport within the aquifer and its supply will run out as the depot becomes completely consolidated. The consolidated situation resembles our experimental simulation. The depot is designed to minimise the hydraulic pressure gradient across the sludge layer. Applying the average measured value for the reflection coefficient of the sample from the Beerkanaal to a sludge layer with an average thickness of 20 m and a hydraulic conductivity of 10–9 m·s–1 —because of a lower consolidation grade than in our experimental samples— yields a water flux of approximately 1.5 mm·y–1. This flux is in the same order of magnitude as the maximum water flux from a sludge in depot of 2 mm·y–1 allowed by Dutch legislation. Therefore, the tentative conclusion can be drawn that a minimisation of the hydraulic gradient, as now is customary to the operation of the depot, not automatically minimises the water flux. Separation of the relatively clean sand and the more contaminated clay fraction before storage in the depot as now is customary, will only enhance the semipermeable properties of the sludge stored. This can result in a larger osmotic water flux than predicted here.

the fate of semipermeable sludge in depot ­­­— In the previous section several assumptions were made, the most important of which the ideality of the sludge layer. Contrary to the assumption, semipermeability is not constant in time. In this research it was shown that the reflection coefficient of one of the harbour sludge samples decreases in time. The reduction of ideality in time was already observed for bentonite samples and originates from the samples' initial non–ideal behaviour. A natural semipermeable material goes through a 'life cycle' starting as a highly but non–ideal membrane. As it is non–ideal diffusion of solutes through the membrane will inevitably render it less ideal with progress of time. Its 'life cycle' will end as the pore water concentration in the membrane is sufficiently high that the semipermeable properties are essentially absent. And given enough time the chemical and hydraulic differences across the ones semipermeable membrane will dissipate. This 'life cycle' was described in the literature for a semipermeable shale in a large geological aquifer setting.

There is a difference between the previous described situation and the possible 'life cycle' of a harbour sludge depot. The following conceptual model involving chemical osmosis for a completely filled sludge depot like De Slufter can be visualised (Figure 2). After completely filling the depot the sludge will go through a consolidation stage in which water is expelled. This is the stage in which contaminants will be mobile and transported out of the depot.

Pasted Graphic

Figure 2
Conceptual model of the stages of a contaminated harbor sludge with semipermeable properties
in a setting similar to De Slufter, the Netherlands (Keijzer, 2000).


The horizontal water flux in the salt water aquifer will dissipate the brackish water lens underneath the depot. Consequently, salt will diffuse upward in the depot, whereas, a hydraulic and osmotic gradient will transport water out of the depot into the aquifer. The water on top of the sludge layer will remain fresh to brackish as a result of rainfall and evaporation. It is in this stage that the initial high σ slowly drops. Given enough time the system will reach a steady–state with the sludge having a low σ. This situation will result in the sludge being capable of acting as a semipermeable membrane over a substantial period of time because of the constant flushing in the aquifer and refreshening of the water on top of the sludge (Figure 2).



concluding remarks


Based on the model assumptions and other theoretical or practical considerations a ranges of conditions and properties given in which the models can be used. Also ways are discussed to tune the models for a closer approach of the conditions encountered in natural materials. It is concluded that both the Fritz-Marine Membrane Model and the Groenevelt-Bolt model can only be used to certain porosity and/or concentration ranges. The Kemper-Bresler model appears to be valid over the entire porosity and concentration range encountered in geological systems. The model predicts relatively low reflection coefficients close to those measured in natural clayey materials in this and other studies.

Based on the observed semipermeability of the Beerkanaal sludge chemical osmosis in a harbour sludge depot like De Slufter is discussed. A conceptual model is used in which the evolution of a sludge membrane is given. It is concluded that the water flux by osmosis can be in the same order of magnitude as the maximum allowed advective flux in Dutch legislation. It is therefore likely that environmental criteria of the depot will be exceeded. Because of the refreshening of the fresh water on top and continues flushing of the salt water aquifer underneath the depot the sludge layer can almost indefinitely act as a semipermeable membrane.

It is currently impossible to estimate or measure the contribution of chemical osmosis on water transport in a field situation. Due to the complex nature of coupled flow phenoma it is my opinion, not possible to determine exactly which part in a given flux is either transported by a hydraulic or a chemical gradient. On a laboratory scale, however, the influences of coupled phenomena on transport mechanisms can be accurately separated and measured as was shown in this and other studies. Experimental work in this area should be done to validate and increase the accuracy of the models, especially when the samples are 'real world mixtures' of clay minerals with different cations on the exchange complex. Experimental work in this field will also increase the awareness that water and solute transport is never governed or by only one driving force.