Journal of Environmental Engineering

This paper reviews the history associated with the occurrence and regulation of disinfection by‐products in finished drinking water, some of the underlying chemistry associated with their formation, technologies and strategies appropriate for their control, and conclusions concerning the current status of their regulation in the United States. Key issues and concerns, along with recommendations for further research, are identified and incorporated into the discussion. The importance of balancing risks associated with disinfection by‐products against risks associated with waterborne pathogenic microorganisms is emphasized, along with the notion that limited information concerning the occurrence and health effects of disinfection by‐products and reliable, safe, and cost‐effective strategies for their control preclude their strict regulation at this time.

During a one‐yr study at Jefferson Parish, La., the chemical, microbiological, and mutagenic effects of using the major drinking water disinfectants (chlorine, chlorine dioxide, chloramine, ozone) were evaluated. Tests were performed on samples collected from various treatment streams. These tests evaluated 18 of the halogenated disinfection by‐products, surrogates such as total organic carbon (TOC) and total organic halide (TOX), microorganisms such as MS2 phage, coliforms, and heterotrophs, assimilable organic carbon, and mutagenicity. The best treatment scheme to control halogenated disinfection by‐products was ozone followed by chloramination. However, the total impact of disinfection has to be evaluated before selecting appropriate disinfectants. There appears to be no disinfectant(s) that is applicable for all situations.

The U.S. Environmental Protection Agency (U.S. EPA) is currently attempting to balance the complex trade‐offs in chemical and microbial risks associated with controlling disinfection and disinfection by‐products (D/DBP) in drinking water. In attempting to achieve this balance, the U.S. EPA will propose three rules: an information collection (ICR); an enhanced surface water treatment rule (ESWTR) and a two‐stage D/DBP rule. Controlling D/DBP will have a major impact on drinking water utilities in the United States. There are several options for D/DBP control, including moving the point of disinfection, removal of by‐products once they are found, removing precursor material or natural organic matter before it interacts with the disinfectant, or use of a disinfectant that minimizes the formation of by‐products. The least‐expensive approach to D/DBP control is to move the point of disinfection or the use of an alternative disinfectant. The least‐desirable approach is to remove disinfection by‐products once they are formed. Overall, the most effective approach to D/DBP control is to remove precursor before it reacts with the disinfectant. The choice of any given strategy is very site specific.

The Safe Drinking Water Act and its Amendments have imposed a large number of new regulations on the U.S. drinking‐water industry. A major set of regulations currently under consideration will control disinfectants and disinfection by‐products. Included in the development of these regulations is an Information Collection Rule and an Enhanced Surface Water Treatment Rule. These rules will require monitoring for microorganisms such as Giardia, Cryptosporidium, and viruses. Certain surface‐water systems may be required to remove microbiological contaminants above levels currently required by the Surface Water Treatment Rule. Also included in these rules will be monitoring requirements for disinfection by‐products and evaluation of precursor removal technologies. As various regulations are promulgated, regulators and those associated with the drinking‐water industry need to be cognizant of the potential impact of treatment to control one contaminant or group of contaminants on control of other contaminants. Compliance with drinking‐water regulations mandated under the Safe Drinking Water Act and its amendments has been estimated to cost about $1.6 billion.

A mass‐transfer‐based model is developed for predicting chlorine decay in drinking‐water distribution networks. The model considers first‐order reactions of chlorine to occur both in the bulk flow and at the pipe wall. The overall rate of the wall reaction is a function of the rate of mass transfer of chlorine to the wall and is therefore dependent on pipe geometry and flow regime. The model can thus explain field observations that show higher chlorine decay rates associated with smaller pipe sizes and higher flow velocities. It has been incorporated into a computer program called EPANET that can perform dynamic water‐quality simulations on complex pipe networks. The model is applied to chlorine measurements taken at nine locations over 53 h from a portion of the South Central Connecticut Regional Water Authority's service area. Good agreement with observed chlorine levels is obtained at locations where the hydraulics are well characterized. The model should prove to be a valuable tool for managing chlorine‐disinfection practices in drinking‐water distribution systems.

Mathematical models previously employed to analyze concentration profiles of dissolved ozone in bubble‐column contactors often assumed either plug flow or complete mixing. In this paper a more realistic model, the axial dispersion model (ADM), was provided to integrate the nonideal mixing, mass‐transfer, and ozone‐decay process as a whole. The effects of various process parameters on the ozone profiles were analyzed numerically. It was found that a group of dimensionless quantities (${\mathrm{Pe}}_L$, $\mathrm{Da}$, ${\mathrm{St}}_G$, and ${\mathrm{St}}_L$) is most convenient in describing the performance of ozone‐contacting systems. The different types of ozone contactors (countercurrent‐flow bubble columns, cocurrent‐flow bubble columns, and reactive contacting chambers) exhibited different patterns of concentration profile; as a result, different approaches are needed to relate their characteristic concentrations to readily observed measurements for predicting the disinfection efficiency. The applicability of the model is illustrated through the analysis of two separate sets of pilot‐scale data. Excellent conformity between the predicted profiles and observations was demonstrated.

Development of models that describe inactivation kinetics with ozone is made difficult by unstable ozone residual, rapid rate of inactivation process, and improper interpretation of inactivation data. Consequently, there are great discrepancies as to both the order of inactivation and the magnitude of reaction rate constants in the rate expressions. In this paper, an innovative completely mixed reactor (continuously stirred tank reactor [CSTR]) was used. Five models, representing the three basic phenomena of inactivation, were fit to the experimental data by applying the rigorous statistical approaches (the classical nonlinear regression or the error‐in‐variables method), where required. It was shown that both the comprehensive model developed herein and the Chick‐Watson model were adequate in predicting the bacterial survival. The empirical coefficient of dilution was found to be as high as 3.3, strongly suggesting that ozone concentration was a much more important factor than contact time in determining the efficiency of the disinfection process. The activation energy of overall inactivation was 85 kJ/mol, indicating that the rate of bacterial inactivation with ozone is chemical reaction‐controlled.

Kinetics of reactions in solutions containing monochloramine and nitrite were evaluated as a function of pH, temperature, excess ammonium, and initial concentrations of monochloramine and nitrite. The ${\mathrm{NH}}_2\mathrm{Cl}$ disappearance rates followed a first‐order reaction with respect to $\left[{\mathrm{NH}}_2\mathrm{Cl}\right]$. The observed reaction rates for nitrite were proportional to $\left[{\mathrm{NO}}_2^{-}\right](1+\mathrm{constant}[{\mathrm{NO}}_2^{-}\left]\right)$. The rate of the reaction was inversely proportional to $[{\mathrm{NH}}_3{]}_T$. The activation energy at pH 7 was relatively low, 13.4 kJ. Phosphate catalyzed the ${\mathrm{NH}}_2\mathrm{Cl}‐{\mathrm{NO}}_2^{-}$ reaction. The analysis of both reactant and by‐product concentrations indicated that nitrogen in ${\mathrm{NO}}_2^{-}$ was not fully oxidized to ${\mathrm{NO}}_3^{-}$, nitrogen in ${\mathrm{NH}}_2\mathrm{C1}$ was not completely converted to ${\mathrm{NH}}_4^{+}$, and the stoichiometric molar ratio of the disappearance of ${\mathrm{NH}}_2\mathrm{Cl}/{\mathrm{NO}}_2^{-}$ was not exactly 1.

Extensive research has shown that microorganisms exhibit increased resistance due to clumping, aggregation, particle association, or modification of antecedent growth conditions. During the course of investigating a major water‐borne Vibrio cholerae outbreak in Peru, U.S. EPA investigators discovered an extremely rough form of Vibrio cholerae that was highly resistance to disinfection by chlorine. This form of Vibrio cholerae also tends to form large aggregates, which contribute to the resistance. Particle counting was used to characterize each of the cultures examined. Statistical analysis revealed that the rugose, or rough, strain of Vibrio cholerae was much more resistant to disinfection than the common, or smooth, strain. Resistance was attributed to both aggregation and a mucoid coating. The larger aggregates associated with the rugose organism are of a size as to be easily removed by filtration, but if the variant were to contaminate a distribution system through a line break or cross connection it would be difficult to control by chlorine disinfection.

A study of biological filtration on granular activated carbon (GAC) has been performed at the Neuilly‐sur‐Marne (France) treatment plant using experimental pilot filters. Bacterial colonization of GAC was monitored in three carbon filters having varying depths and filtration velocities, but similar empty bed contact time (EBCT). About three months were required to colonize the GAC in the filters and a steady state bacterial biomass of around 2 μgC/cm³ GAC was reached. During the colonization period, removal of nonbiodegradable organic carbon (NBDOC) by filtration decreased from 50% to less than 10% showing a gradual saturation of GAC adsorption capacity. However, biological processes take turn with adsorption to ensure a significant removal of biodegradable dissolved organic carbon (BDOC) during the study period (more than 200 days). Filters operating under different conditions did not show any difference in the BDOC removal. This indicates that biological removal of organic matter in GAC filters having a given EBCT is independent of filtration rate (in the range 6 m/h–18 m/h). The maximum bacterial density in the filter effluent water was $1.5\times {10}^5\mathrm{bacteria}/\mathrm{mL}$ during the colonization phase; this seemed to be linked to a lag in establishing a protozoan population, which, at equilibrium, efficiently removed bacterial biomass produced within the filters. Once established, the efficiency of this complex microbial assemblage was remarkable; the removal of about 0.25 mgC/L of BDOC resulted in the release of less than $0.5\times {10}^5\mathrm{bacteria}\mathrm{per}\mathrm{millileter}$, around 0.0015 mgC/L as bacterial biomass in the effluent.

Although a number of optimization models have been developed for the least‐cost design of wastewater‐treatment systems, very few models are available for water‐treatment systems. The present study proposes a method toward optimal design of water‐treatment systems, considering integration of process optimization in system optimization; particle‐size distribution (PSD) for process design and selection; relatively general simulation models; membrane processes in the alternative evaluation; sludge treatment/handling processes; multicriteria for alternative evaluation; and algorithms for process and system optimization. The paper is divided into two parts. The first part presents the important features of simulation models along with the simulation results and the second part highlights the optimization models and algorithms used for the optimal design. Results from the simulation study indicate the importance of influent particle‐size distribution on the performance of flocculation, sedimentation, and filtration processes. Recycling of the sludge sidestreams from gravity/centrifugal separation processes to the head of the water‐treatment plant and the filtrate disposal into sewers are economically and environmentally justified.

In this second part of the paper, optimization models for design of water treatment plants for turbidity removal are discussed. Application of the models is illustrated with a case study. The results from the case study indicate cost differences of about 16–64%, solely due to change in the influent particle size distribution (PSD), with higher cost of treatment for the influents with relatively greater fraction of fine particles. The system configuration, level of treatment, and design and operating parameters are found to be different for each influent PSD. Use of other parameters like turbidity units, suspended solids and volume average diameter are unable to explain this variation. The influence of PSD on the total cost decreases with inflow. Integration of process optimization in system optimization is justified as the different design and operating parameters are selected under each level of treatment. The cost of sludge treatment accounts to about 30–50% of the total plant cost.

A determining factor for the performance of secondary settling tanks is the occurrence of gravity currents. Experiments were carried out in a laboratory model of a circular, center‐feed settling tank. The gravity currents were scaled using the internal Froude number criterion, and the settling velocity was scaled on the basis of kinematic similarity. The activated sludge was represented by kaolinite. A flow pattern with three layers of different densities and different flow directions was observed. A simple one‐dimensional analytical model taking into account the influence of flocculation was developed to predict the effluent concentration. It was found that gravity currents may result in a layered flow pattern reducing short circuiting. This flow pattern thus seems to promote settling rather than to deteriorate it as is generally assumed.

The prediction of head loss and filtrate concentration are of considerable interest to the users of deep‐bed filters. At present, such prediction is done by using existing theoretical models. However, it is also known that the coefficients in these theoretical models are sensitive to changing filtration conditions. Unless a wide range of possible conditions (which occur in a filter run) have been investigated through trial filter runs, predictions using conventional models may not be always reliable. In such circumstances, the use of recursive algorithms, which represent an interrelationship between various equally spaced sequential outcomes, may be very useful. It is possible that very simple recurrence relationships are capable of describing filter performance satisfactorily. In this study, the potential use of recursive algorithms in predicting filter performance has been examined through the analysis of several laboratory and field‐based pilot‐scale filter runs.

A method of designing least‐cost branched‐water‐supply networks for rural communities on uneven terrain is described. Previous studies have been concerned with networks on relatively flat terrain and have not taken account of the need to provide minimum‐service pressure heads at interior nodes. The optimization model requires a function representing the consolidated cost of the pipe to be minimized subject to a number of constraints on the friction head loss to be tolerated between the source node and a critical node. The critical node may be an end node or an internal node. Optimization is achieved using the Lagrange multiplier method, which gives rise to as many nonlinear simultaneous equations as there are critical nodes. The equations for the multipliers are solved using a generalized Newton‐Raphson approach, which is rapidly convergent. The method is applied to a small‐scale continuous system typical of those found in Papua New Guinea.