Grantee Research Project Results
Final Report: Investigation and Optimization of Dual Coagulation Processes
EPA Grant Number: R822462Title: Investigation and Optimization of Dual Coagulation Processes
Investigators: Benjamin, Mark M. , Edwards, Marc
Institution: University of Washington , University of Colorado at Boulder
EPA Project Officer: Aja, Hayley
Project Period: October 1, 1995 through September 30, 1998 (Extended to April 30, 1999)
Project Amount: $273,209
RFA: Exploratory Research - Engineering (1995) RFA Text | Recipients Lists
Research Category: Safer Chemicals , Land and Waste Management
Objective:
Historically, coagulation and flocculation have been used in drinking water treatment for the removal of particulates. The Enhanced Coagulation provision of the proposed Disinfectants/Disinfection By-Products (D/DBP) Rule has changed this situation by mandating that potable water treatment processes be operated to achieve specified removal efficiencies of total organic carbon (TOC). This project investigated the effectiveness of combinations of organic and inorganic coagulants on natural organic matter (NOM) removal from potable water sources. The project was carried out jointly by researchers at the University of Colorado at Boulder, who developed a model for NOM coagulation by ferric chloride or alum, and others at the University of Washington (UW), whose work focused on NOM removal by combinations of ferric chloride and organic polymers.Summary/Accomplishments (Outputs/Outcomes):
In the work at UW, surface water was dosed with varying concentrations of ferric chloride and cationic or anionic polymers. Particle zeta potential was measured immediately, and the turbidity and dissolved organic carbon (DOC) concentration of the water after settling were analyzed. In some cases, the turbidity of the raw water was increased by addition of bentonite.When ferric chloride was added as the sole coagulant, increasing the coagulant dose increased the zeta potential of the particles, but the post-settling turbidity often increased with increasing coagulant dose at low doses. At higher Fe doses, the turbidity decreased substantially. The DOC removal efficiency increased with increasing coagulant at low coagulant doses, but at higher doses, it usually reached a plateau. The probable explanation for this is that Fe(OH)3 particles that were formed by precipitation of the coagulant were stabilized by the DOC and, hence, did not settle well. In the absence of other particles on which to sorb (or with which to coagulate), these stable Fe(OH)3-NOM colloids contributed to turbidity. No consistent correlation could be found between the zeta potential and the formation of settleable flocs. Although there was a good deal of scatter in the data, the zeta potential and the DOC removal efficiency both generally decreased with increasing pH; however, the turbidity of the settled water was insensitive to pH.
In water that had been amended with bentonite, the turbidity of the settled water declined monotonically with increasing coagulant dose. Apparently, in this system, the initial particle concentration was sufficiently high that, at low FeCl3 doses, all of the added Fe sorbed. This sorbed Fe decreased the negative charge on the particles and enhanced their removal.
Addition of the cationic polymer Cat-Floc T as a sole coagulant removed only a small amount of DOC and increased the turbidity of the settled water. Therefore, using (this) cationic polymer as primary coagulant does not appear to offer any benefits. When Cat-Floc T was added in conjunction with ferric chloride, the polymer enhanced DOC removal slightly at low Fe doses and had a negligible effect on DOC at higher doses.
When an anionic polymer was used as the sole coagulant, it did not cause any visible flocculation, regardless of the water's original turbidity, and the polymer itself contributed to the DOC of the water. As a result, the treated water had a higher DOC than the raw water. In systems dosed with both FeCl3 and anionic polymer, the polymer had an extremely minimal effect on the zeta potential of the particles and on the DOC removal efficiency, respectively, in these experiments. On the other hand, the polymer enhanced the removal of turbidity from the suspension quite dramatically. The effectiveness of the anionic polymer at lowering the settled water turbidity, even when the overall charge on the colloids was negative, can be attributed to the formation of positively charged ferric hydroxide colloids, once some critical Fe dose had been added. The anionic polymer had a minimal effect on the particles' zeta potential at any pH and Fe dose. Anionic polymers probably will be useful in enhanced coagulation only in those relatively rare cases where the turbidity removal is inadequate after sufficient primary coagulant has been added to achieve the DOC removal goal.
The zeta potential in systems where the polymer was added prior to FeCl3 was consistently and significantly lower (more negative or less positive) than when the FeCl3 was added first, even though the same doses of FeCl3 and polymer were used. The turbidity of the settled water in the systems with polymer added before FeCl3 usually was very close to that in the Fe-only systems and lower than in the systems where FeCl3 was added prior to the polymer. Removal of total organic carbon (TOC) was approximately the same or slightly better when the coagulants were added in the conventional order (FeCl3 before polymer) rather than the reverse order.
A model for TOC removal by enhanced coagulation processes was developed at the University of Colorado. This model assumes that the TOC in a water supply can be operationally divided into three sub-groups?a soluble, adsorbable fraction; a soluble, non-sorbable fraction; and a particulate fraction. Interaction of the soluble, adsorbable fraction with added coagulant is modeled by the Langmuir adsorption isotherm, with pH- and water-source-dependent values for the value of the Langmuir affinity constant. The non-adsorbable fraction (i.e., the TOC that does not sorb at any practical coagulant dose) can be related to the specific UV absorbance of the water and can range from almost none to approximately one-half of the DOC, depending on the water source. This DOC is presumed to remain in solution under all coagulation conditions of interest, while the particulate fraction is presumed to be removed under all such conditions.
An important aspect of the model application is the estimation of solution pH after coagulant addition. The calculation of solution pH is based on the conservation of alkalinity, considering the alkalinity of the original water and the added coagulant (including any acid or base in the concentrated coagulant solution, beyond that contributed by the metal salts themselves) and assuming that all of the coagulant metal precipitates as Me(OH)3(s), where Me is either Fe or Al, depending on the coagulant identity. In addition, the fate of carbonate species in the system is taken into account by treating the system as either open or closed to the atmosphere, depending on the details of the physical setup. While these calculations are a straightforward application of water chemistry principles, the important contributions of this project are the recognition that CO2 exchange with the atmosphere accompanying the rapid and slow mixing steps can have a significant effect on the pH of the coagulated solution, and that this effect can be significantly different in jar tests from that in the full-scale system, because of the greater opportunity for CO2 exchange in the former case.
The model was used to estimate TOC removal by enhanced coagulation with both Al- and Fe-based coagulants at 27 full-scale systems. The conditions investigated included various blends of two source waters, in which case the prediction of TOC removal was based solely on data from jar tests using the two unblended end members. In those cases, the pH of the blended, coagulated water was predicted as described above, and the overall removal of sorbable TOC was based on competitive Langmuir adsorption of the sorbable TOC from the two end members.
The results of the modeling were very encouraging. The Langmuir affinity parameter for TOC sorption was source-dependent, but was, to a good approximation, independent of solution pH. The maximum adsorption density (qmax) decreased in a consistent way with pH and was slightly greater for sorption onto Fe(OH)3 than for sorption onto Al(OH)3. For both solids, this dependence could be modeled by an equation of the form qmax=x1(pH)3+x2(pH)2+x3(pH), where the parameters x1, x2, and x3 are specific to the source water being investigated. However, once these parameters are calibrated to a particular source water, they can be used to provide good predictions of DOC removal over time, even if the DOC concentration in the raw water changes substantially. In most cases, the predictions for DOC remaining after enhanced coagulation were within 0.25 mg/L of the observed value.
Modeling of the effect of coagulation on solution pH also was successful. Significantly, the presumption that full-scale systems are, to a good approximation, closed with respect to gas exchange with the atmosphere, while solutions used in jar tests are open to the atmosphere, led to predictions that the pH change in full-scale systems would be different from that in jar tests. In turn, these differences could lead to significant differences in TOC removal in the two types of systems. These differences were confirmed experimentally, suggesting that, in some cases, the model was a better predictor of full-scale performance than jar tests.
When the model was adapted to predict TOC removal in systems where two raw waters are mixed in varying proportions, the standard error was only 0.15 mg/L for 2 years of jar test data at one utility. In case studies of 27 full-scale utilities, accurate prediction of TOC removal by coagulation was illustrated at a range of utilities using alum, ferric, or polyaluminum chloride coagulants; over long time periods without a need for frequent re-calibration; using a predicted rather than measured coagulation pH; and for two raw waters mixed in various proportions.
Conclusions:
Qualitatively, these are changes in particle zeta potential when a suspension is dosed with a cationic or anionic coagulant; however, the effects of these changes on NOM removal and particle behavior often are not explainable by consideration of electrostatic effects alone. Cationic and anionic polymers (at least those studied in the current project) can have a dramatic effect on particle-particle interactions (coagulation) at low doses; however, their incremental effects tend to be quite small at higher doses, even if they affect the zeta potential substantially. They tend to have small effects on NOM removal at any dose. Ferric chloride, on the other hand, enhances coagulation more steadily over an extended range of doses. It also adsorbs or precipitates steadily more NOM with increasing dose, until all the removable NOM in the water (typically 60 to 90 percent of the total NOM) has been insolubilized. Consistent with the dosing schemes that have developed empirically over the years at water treatment plants, combinations of conventional doses of ferric chloride and small doses of polymer appear to offer the greatest advantage in terms of combined particle and NOM removal. The NOM removal can be modeled well by treating the Fe(OH)3 (or Al(OH)3) that precipitates upon ferric chloride addition as a conventional adsorbent, and modeling the adsorption reaction with the Langmuir isotherm. In doing so, accounting for pH changes in response to the precipitation and adsorption reactions, as well as CO2 exchange with the atmosphere and any other acid/base reactions in solution, can be critical.Journal Articles on this Report : 2 Displayed | Download in RIS Format
Other project views: | All 3 publications | 2 publications in selected types | All 2 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Edwards M. Predicting DOC removal during enhanced coagulation. Journal of the American Water Works Association 1997;89(5):78-89. |
R822462 (Final) |
not available |
|
Tseng T, Edwards M. Predicting full-scale TOC removal. Journal of the American Water Works Association 1999;91(4):159-170. |
R822462 (Final) |
not available |
Supplemental Keywords:
water, drinking water, water treatment, dual coagulation, modeling, efficiency prediction., RFA, Scientific Discipline, Water, Environmental Chemistry, Chemistry, Drinking Water, Engineering, alternative disinfection methods, aluminum-based coagulants, treatability of flocs, water quality parameters, flocculation, efficiency predicition, water utilities, treatment, environmental engineering, iron, polymers, DBP risk management, drinking water contaminants, drinking water treatment, dual coagulation process, regulations, metal coagulation, other - risk management, drinking water systemProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.