Quantification of factors affecting coagulation of water with cationic polymers and laboratory methods for determining these effects.
January 2004

Report No 1225/1/04

Executive Summary

Coagulation is one of the most important aspects of potable water treatment, being essential in the separation of solids and providing a primary barrier against waterborne diseases. Iron and aluminium salts are often used as primary coagulants and the reactions that occur with these coagulants are fairly well elucidated. More recently, organic polyelectrolyte coagulants have become more widely used, but the reactions of these chemicals are not as well understood as those for their inorganic counterparts.

Anomalies have been observed in Umgeni Water's operational area, which complicate coagulant selection and dose optimisation. For example, augmentation of uMngeni River water in Midmar Dam with water from the Mooi River results in a significant change in coagulant dose and the type of coagulant best suited for the treatment of the water changes, despite the fact that no noticeable changes to the obvious water quality parameters occur and that the volume of Mooi River water added to Midmar Dam has been relatively small.

Tests have been conducted on water samples from three areas where anomalies have been observed, namely the Midmar/Mearns system, the Durban Heights/Amanzimtoti/Nungwane system in the greater Durban area and the Mvoti/Makovane system on the KwaZulu-Natal North Coast, but on the advice of the Steering Committee, testing was concentrated on the Midmar/Mearns system.

In addition to this, the evaluation of operational data acquired over the years by Umgeni Water was carried out. Certain sample points within the Umgeni Water Operational area have been monitored for a number of years and this data has been analysed in order to assist in identifying factors which are important in terms of coagulation. Data were analysed for the Midmar Dam raw water and Mearns sampling sites from the Midmar-Mearns system. In the case of the Midmar Dam raw water, this data includes the coagulant type and dose being used as well as a number of water quality parameters and the flow rates of the Midmar and Mearns water into Midmar Dam.

The jar test, which has always been used successfully for dose selection when using inorganic coagulants, is often inadequate for coagulant type and dose selection when using polyelectrolytes. Modifications to the jar test are described which improve correlation between this test and full-scale operation. This was carried out in order to address the aim of the second research product and to meet the third objective if this investigation.

A better understanding of the factors affecting coagulation with organic polyelectrolytes would allow for more rapid and accurate selection of the correct type of polyelectrolyte and dose. This investigation was conducted in an attempt to provide the answer to some of these questions.

1.1. Project Objectives

The objections of this project as specified in the original project proposal are as follows:

  1. Elucidate the chemical reactions that occur during coagulation using polyelectrolytes.
  2. Characterise Southern African waters in order to determine the effect of natural organic matter on polyelectrolyte coagulants.
  3. Produce procedures and tests to enable accurate and easy selection of polyelectrolyte coagulant type and dose for a particular water type.

The two predominant research products that the researchers hoped to produce from this project were:

  1. Assessment of the effect of natural organic matter (NOM) on coagulation when using polyelectrolyte coagulants.
  2. Procedures for the rapid and accurate selection of polyelectrolyte type and dose.

In conjunction with the laboratory tests conducted for this project, an in-depth data analysis was conducted on a large database of historical data, including both water quality and operational data. The objectives of this data analysis and interpretation were to assess:

  1. Differences in land cover and water quality in the upper Mooi and upper Mngeni catchments that will provide an indication of the cause of the Waterworks (WW) coagulant dose during transfer periods.
  2. Historical WW coagulant dose during transfer and non-transfer periods.
  3. The relationship between coagulant dose and selected water quality constituents to assist in predicting changes in coagulant dose during transfer periods.

1.2. Results and Discussion

At the start of this project, systems where anomalies in coagulation existed were identified and the raw waters from these systems, as well as mixtures of these waters were assessed in terms of optimal polyelectrolyte coagulant dose and most suitable coagulant, with special emphasis being placed on the impact that mixing of the different waters has on both factors. The systems chosen were the Mvoti-Makovane system, the Durban Heights- Amanzimtoti-Nungwane system and the Midmar-Mearns system.

The Mvoti-Makovane system is being planned for water storage in the Stanger area on the KwaZulu-Natal North Coast. The Durban-Amanzimtoti-Nungwane system includes the Durban Heights and Amanzimtoti Water Works as well as water from the Nungwane Dam and although all three of these waters come from different sources, they are geographically all within fairly close proximity on the coast of KwaZulu-Natal and in terms of water quality parameters are very similar. Despite this, these waters respond very differently to polyelectrolye coagulants.

The Midmar Dam supplies the greater Pietermaritzburg area with water and on account of its strategic importance, an augmentation scheme was commissioned in 1983 as an emergency measure during the drought experienced in the uMngeni catchment. This scheme allows water from the Mooi River at Mearns to be pumped into the Lions River which in turn flows into the uMngeni River shortly before it enters Midmar Dam. It has been observed that whenever water from the Mooi River has been used to augment Midmar Dam, the water responds very differently to coagulation when treated at the nearby Midmar Water Works, despite no obvious changes in the water quality of the raw water and the fact that the Mooi River water accounts for only a small proportion of the total (less than 10%). At the start of this project in 2001, water from the planned Springrove impoundment area were included in the tests, but in later tests, the Springrove water was excluded as the Springrove development has been placed on hold indefinitely. The later tests conducted after mid 2001 included waters and blends of waters from only the uMngeni River, both above and below the confluence of the Lions River, Midmar Dam itself and the Mearns weir, where water is taken for augmentation of Midmar Dam.

1.2.1. Methodology

The tests on initially the three water systems and then later on only the Midmar-Mearns system, were conducted at laboratory scale using jar tests and although it had originally been planned to conduct pilot plant tests as well, once the results of the laboratory tests were known, the pilot-plant tests were abandoned.

Standard jar tests were performed on each raw water source and, where relevant, any blends of these waters, using a range of polyelectrolytes and aluminium sulphate. Tests to assess variations in coagulant demand were conducted using polyelectrolytes which were chosen to cover the variety currently available on the Southern African market, namely:

  1. A polyamine (PA)
  2. A dimethyldiallyl ammonium chloride (DMDAAC)
  3. A blended PA and polyaluminium chloride (PACl)
  4. A blended DMDAAC and PACl.

Aluminium sulphate, an inorganic coagulant, was used in these tests.

Comprehensive analysis of the various water quality parameters was carried out together with characterisation of the natural organic matter present in the water. The analyses used to assess general water quality of the water samples both before and after treatment as well as before and after blending, included the following:

  1. turbidity
  2. pH
  3. alkalinity
  4. calcium, magnesium, hardness
  5. colour
  6. conductivity
  7. iron, manganese (total)
  8. suspended solids
  9. total dissolved solids

In order to achieve the first research product of this study, characterisation of the natural organic matter (NOM) present in the water was done by analysing for the following;

  1. total and dissolved organic carbon (TOC and DOC)
  2. biodegradable dissolved organic carbon (BDOC)
  3. trihalomethane formation potential (THMFP)
  4. absorbance at 254 nm
  5. chlorine demand
  6. lime demand
  7. zeta potential
  8. gas chromatograph-mass spectrometry (GC-MS) fingerprinting

"Titration" curves were obtained for various blends of water samples taken from the Midmar-Mearns system, in which incremental amounts of one water sample would be added to another, until a 1:1 blend had been achieved. After each incremental addition, the turbidity, pH, conductivity and zeta potential were measured.

Tests were conducted on organic polymeric coagulants which varied in molecular mass, charge density and constituents in an attempt to determine the impact of these factors in the coagulant reaction. A variety of laboratory jar tests were conducted using a range of these coagulants and again the determinands described above were analysed.

Enhanced coagulation tests were conducted on water from the Midmar-Mearns system with a view to identifying differences within the organic constituents of the different waters. Ozonation of the various waters from the Midmar - Mearns system was carried out in order to identify any differences in the NOM present in these waters.

Tests to assess the effect of removal of the particulate matter prior to coagulation were conducted by filtering the individual waters and blends with GF/C filters (1,2 Ám) and Whatman No. 1 equivalent filter paper before coagulant addition.

In addition to the laboratory tests which were conducted, a detailed data analysis and interpretation study was conducted. A large database of historical data, including both water quality and operational data were used for this purpose.

Laboratory tests were carried out in an attempt to improve the correlation between the jar test results and full-scale operation. These included laboratory-scale clarifiers as well as filtration tests conducted in conjunction with full-scale operation.

1.3. Summary of the Results

As mentioned above, three systems were originally included in the investigation. These were the Mvoti/Makovane system near Stanger, the Durban Heights/ Nungwane system near Amanzimtoti and the Midmar/Mearns system Inland of Pietermaritizburg. A first set of tests was carried out on all three systems and showed a confirmation of the dosage anomalies previously noted.

1.3.1. Mvoti/Makovane System

The first set of results taken confirmed the anomalies previously noted, but results tended to be somewhat unpredictable. As the system did not have a large database of previous results and was remote from the laboratories, leading to practical difficulties in sampling and logistics, it was agreed at the first steering committee to drop this system from the list and no further investigations were carried out.

1.3.2. Durban Heights/Nungwane System

The second system investigated, comprising Durban Heights water and water from Nungwane Dam treated at the Amanzimtoti works, had a certain amount of previous data and works records which clearly showed anomalies in the coagulant demands between the Durban Heights and Nungwane water despite superficially similar physical and chemical characteristics. These anomalies were also apparent in the first set of tests carried out. However, when the amount of data processing to produce meaningful results from the historical data became apparent, it was evident that it would not have been possible to fully investigate more than one system. The decision was therefore taken at the steering committee to limit work to the Midmar/Mearns System which was closest to the laboratories, and had by far the largest base of historical data.

1.3.3. Midmar/Mearns System

After the first set of investigational tests, work concentrated on the Midmar/Mearns System. As a first step and in parallel with the investigation on parameters which were not normally analysed routinely, a detailed analysis of all historical data was carried out on water treated in the Midmar system. Samples taken over a five-year period on a weekly basis were analysed for correlation between coagulant demand and the various parameters normally measured. These data are presented in the body of the report as Chapter 4 and summarises the problem experienced. In this study an attempt was made to correlate coagulant demand with all routinely measured variables. Correlation coefficients were calculated between each variable and the correlation coefficients are presented in the body of the report. Virtually no correlation was evident between any of the variables measured although a weak correlation (correlation coefficients between 0,25 and 0,3) was found for several parameters including TDS (but not Conductivity), Sodium, Barium, Colour and Nitrate.

It was considered that the correlation for Barium was fortuitous and this aspect was not pursued as the concentrations under consideration were very low and only partial data were available. A weak correlation between turbidity and coagulant demand was expected based on previous experience but this was not established. It had been noted on previous routine samples that a weak correlation exists between dissolved solids content and coagulant demand. The correlation with sodium is an echo of this relationship in that the sodium content would be expected to increase with TDS. No correlation was found to exist, with the limited work done, between organic content of the water although the correlation between nitrate was indirectly indicative of possible organic enrichment. This work was then expanded in the investigational work carried out and reported subsequently.

1.3.4. Midmar/Mearns Experimental Work

Tests carried out over the period of the project, which encompassed two years and therefore two full seasonal variations, showed confirmation that the presence of Mearns water had a disproportionate effect on the coagulant demand when mixed with water from the uMngeni River and Lions River which is the normal supply to Midmar Dam. This however was expected as it was the observation which led to motivation for the investigation in the first place. It was also confirmed that the effect of Mearns water on the coagulant demand was stronger then would be expected in proportion to the amount of water present in the various blends.

It had been anticipated that the difference in coagulant demand may have been due to organic content which is not normally measured in routine testing. The characterisation of natural organic material (NOM) was measured in this investigation in a number of ways. It was found with the experimental work that generally no correlation existed between the organic surrogates and coagulant demand. Although a weak correlation existed between UV absorption and turbidity, this was insufficient to be significant. The second objective, which would have resulted in the first research product of this investigation was not then fully realised.

The investigational work was largely confined to the use of polyelectrolytes for coagulation as these tended to display a greater anomaly in demand compared to the inorganic coagulants such as aluminium sulphate. No significant correlation existed between the molecular mass and charge density of the polyelectrolytes, or whether these consisted of a DIMDAAC or a polyamine in origin and the coagulant demand. To further explore this, special samples were obtained from one of the chemical suppliers who produced a range of polymers for our purposes consisting of the same chemical but having different molecular masses and surface charges ranging from very low molecular mass to a very high molecular mass with accompanying variation in surface charge. Again no significant effects or correlations were noted although a higher coagulant dose was evident with the low molecular mass product, this did not vary significantly between the different samples analysed. Thus there were no marked changes in reactions obtained with different formulations of polymers. This had been intended for investigation as part of the first objective of this study.

It was decided to test the waters using dosages in the enhanced coagulation range to see whether the anomalies persisted at higher organic compound removals. Aluminum sulphate was used for this purpose as restabilisation of polyeletrolytes occurs before the enhanced coagulation effect becomes apparent. Again in this case no trends were noted and results tended to echo the results obtained at the normal dosages for turbidity removal.

It had been hoped that the availability of a zeta potential meter and streaming current analyser would assist in characterising the waters in such a way that some explanation for the anomalies related to surface charge could be provided. Zeta potential measurements were carried out on all samples tested from the date the meter was received and no significant correlation was obtained with any of the samples. All raw water samples tended to have similar zeta potentials prior to addition of coagulant and similar zeta potentials at the point of optimum turbidity removal. There was no difference between the zeta potentials of the different water samples that could significantly account for the differences in coagulant demand. The results with the streaming current analyser were almost identical to those obtained using the zeta potential meter, but this is to be expected as they measure the same effect, one in terms of potential and one in terms of current. The difference in either the zeta potential or streaming current detector (SCD) measurement of the raw water and the treated water appears to be related to the coagulant rather than the water.

Ozonation of samples was carried out to check whether the modification of the organic species in the samples would affect the coagulant demand. Again, no significant differences between the different types of water were noted.

It had been postulated by other researchers in the field that coagulant demand for polyelectrolytes was governed by organic rather than inorganic suspended solids, and at the suggestion of the steering committee an additional series of tests was carried out where the raw water was filtered to remove TOC. By doing this it was hoped that a correlation between coagulant demand and TOC removal could be identified. The results however indicated a greater effect by filtration on inorganic matter than on TOC, as nearly all the organic carbon appeared to be in the dissolved form, and no readily identifiable correlations occurred.

GC-MS screening of the different raw water samples was carried out in order to ascertain whether this might have yielded a reason for the difference in coagulant demands. The scanning curves are presented in the body of the report and it was apparent on examination of these that all the peaks could be adequately accounted for by impurities in the solvent used to extract the samples.

1.4. Conclusions

The general conclusion which can be drawn from all the work carried out in this investigation is that the tools used to measure differences in water quality for correlation with coagulant demand were not significant in their effects as far as predicting the coagulant demand is concerned. It had been hoped that measuring of the organic species or the zeta potential and streaming current would yield a reason why the coagulant demands varied. But this was not apparent in the tests. It can therefore be concluded that more detailed work would need to be carried out into fundamental characteristics of the particles in suspension to possibly account for the differences noted in coagulant demand. Thus the third objective and the second research product were not fully realised and achieved.

1.5. Recommendations for Future Work

As mentioned in the conclusions the parameters measured did not significantly account for the differences noted in coagulant demand. Future research could possibly take into account molecular formulae, crystal structures and charges on the inorganic particles present in suspension in the water and could also be extended to other water systems where these effects have been noted. The catchments could be studied to note whether there are any differences in geology of the two catchments which could yield dissimilar particles which could give rise to the differences in coagulant demand.

Another possibility would be to examine in detail the spread of algal populations in the water and to establish whether any correlation exists between a particular algal species and coagulant demand.

A third possibility would be to investigate the nature of the organic material present in the water in solution or in colloidal form, such as humic substances and attempt to correlate these with coagulant demand in some way. It was mentioned in the report that fractionation of the organics measured by TOC and BDOC would not have yielded anything significant, but it is possible that concentration of the organics or more sensitive analysis of the organics present might yield information of interest.