Samenvatting

Drinking water utilities are responsible for providing safe drinking water. As water directly impacts public health, water utilities follow stringent water treatment standards and processes. However to keep pace with the dynamic environment and tackle challenges such as water stress, water pollution, climate change, and sustainability, water utilities are required to adapt their processes. In order to adapt, careful research is done beforehand on its consequences. Amongst others, prediction models are used to foresee these consequences.

Waternet is a water utility in Amsterdam (The Netherlands) that uses natural grains in their
multiphase flow processes such as pellet-softening, slow sand filtration and rapid sand filtration. While these natural grains are non-spherical in shape, most prediction models assume the input of perfect spheres and correct by applying a particle shape correction factor. This research aims to focus on how to account for non-spherical particles.

Literature investigation is performed on how to account for non-spherical particles in a voidage prediction model which showed that no general agreement or consensus is illustrated in the literature accounting for particle shapes and size in a voidage prediction model. Operational field of water treatment lies in the vicinity of incipient fluidisation and laminar/transitional regime is preferred therefore, a conventional model such as Carman-Kozeny is used, and the new empirical data-driven model to find out the estimated spherical diameter for non-spherical particles by using the experimental data collected from expansion experiments.

All lab experiments for spherical glass beads (0.8-3.5 mm) and non-spherical glass rods (3x9 mm) were conducted at a wide range of 5-35 degrees Celsius, simulating the seasonal water temperature range at Waternet and beyond. The results of the experimental work have shown that particles used by the treatment plants are dependent on flow and it has an effect on the expansion behaviour which can effect crystallisation process in pellet softening process.
Nevertheless, spherical particles are independent of flow. Furthermore, it has been shown that the new empirical data-driven model which is applicable for spherical particles does not work well for non-spherical particles. The present research has also focused on the orientation of the non-spherical particles which change with fluid flow rate. The dimensionless number approach has been applied to experimental data of non-spherical glass beads using Carman Kozeny drag relations. It has been found that the Carman-Kozeny constant is not constant in a fixed and fluidised bed state. Mostly, the shape factor is applied only in the fixed bed state but in the fluidised state it is simply omitted. Nevertheless, with the help of experimental data, the dynamic shape factor has now been determined in the fixed as well as in the fluidised state.