Banner - Hydraulic Research and Testing Facility
Research Opportunities
Dynamic Behavior of Head Loss
in a Deep Bed Granular Filter Based on
Deposition Morphology

Al Beldawi Mohammad Fadhil

A review of the literature related to assessing treatment efficiency and head loss build-up using particle count and size distribution indicates some deficiencies and limitations. These deficiencies and limitations indicate several shortcomings in the available head loss models. It was noted that previous research work, which used particle size analyzers, have experienced difficulty and uncertainty in obtaining the true particle count. A procedure to correct the particle count, supported by quantitative and qualitative validation, is provided. Result obtained from past investigations on floc geometry, and its mode of deposition, have also displayed a pronounced disparity. Furthermore, no photographic evidence has been published that shows the particle dendrites in liquid. Most observations of deposition patterns relate to conditions that exhibit a substantially different behaviour from conventional filter media characteristics.

The objective of this work was to develop a mathematical model which describes the headloss in deep bed granular filter based on different modes of deposition. The deposition models include parameters that account for the dynamic fraction of particles that contribute to additional surface area and the change in geometry. The dynamic fraction of particles that contribute to smooth coating and dendrite deposition in filter containing deposit are also included in the model. The deposition morphology was validated visually using an optical fibre endoscope and quantitatively using the filtration test.

Three mathematical models have been developed for describing the effect of different deposit morphologies on the head loss build-up in sand filters. The models predict the head loss for a specific deposition mode and a combination of (different) modes. The first model is based on the assumption that retained particles form a relatively smooth coating around the filter grains (smooth coating model). The second model is based on the assumption that retained particles can act as additional collectors, forming chain-like depositions (dendrites mode model) as proposed by O'Melia and Ali (1978). A refinement to the O'Melia model is also proposed. In particular, the variation in porosity and the factor which represents the fraction of retained particles that contribute to the additional surface area were considered in the model refinement. The effect of change on the geometry of the filter grains were considered as well. In the third model, adsorption and bridging were proposed as combined mechanisms of polyelectrolyte action during filtration of small silica (5 um) suspension. Therefore, this model is based on the assumption that the deposition takes place in two stages: smooth coating and dendrite deposition (combined mode model). It is also hypothesized that more than one mode of deposition may govern the deposition process during the filter time.

A comprehensive experimental program was conducted in which the size and concentration of silica suspension, as well as the filtration velocity and type of coagulant, were varied while the filter column and media characteristics were held constant.

Results from the experimental program and analytical modelling were used to validate the proposed deposition morphologies and to map-out the important factors responsible for different geometries and their contribution to head loss build-up. Reasonable agreement was observed between the measured and predicted head loss. The proposed models are general in nature and may be extended to conditions other than those used for validation, provided that the deposit morphology is the same. The accuracy of the proposed combination of different deposit morphologies was confirmed using experimental data.

An analysis and possible explanation for the similarity and dissimilarity of the optimum coagulant dosage between the results from a jar test and filtration test are presented.

| Return to Research |