Enhancing climate adaptation capacity for drinking water treatment facilities

Water supplies are vulnerable to a host of climate-relevant stressors such as droughts, intense storms/flooding, snowpack depletion, storm surge, sea level changes, and consequences from fires, landslides, and excessive heat or cold spells (Intergovernmental Panel on Climate Change (IPCC) 2014). Given that there are over 150,000 public water systems in the USA (http://water.epa.gov/infrastructure/drinkingwater/pws/factoids.cfm) that deliver drinking water to over 300 million people every day, it is important to evaluate the adaptation capacity and resilience of these systems. Resilience can be defined as the ability to physically provide, repair, and recover its service functions following a disruption (McDaniels et al. 2008; Milman & Short 2008). In the context of water systems, resilience encompasses the source water and the infrastructure used to treat and convey water to and from end-users. According to the National Infrastructure Advisory Council (NIAC) (2009):

Surface water resources (lakes, reservoirs, rivers, and streams) provide drinking water to approximately 70% of the US population, and are especially susceptible to extreme weather and climate-induced changes in water availability and water quality. Some climate and related meteorological events trigger abrupt water quality disturbances, such as erosion-induced increases in sediment and runoff following intense storms and flooding. Transient water quality impairments may also occur in the aftermath of spills, fires, accidental or intentional chemical releases, or security breaches (e.g., Ruhl et al. 2012; Bladon et al. 2014; Gallagher et al. 2015; Whelton et al. 2015). Climate-enhanced chronic droughts and their intensity can also directly or indirectly affect water quality due to changes in evaporation rates, temperature, salinity, aquatic habitats, and available surface area for gas exchange. Longer-term systemic impacts include changes in water availability, the frequency and intensity of algal blooms, gradual changes in the nature and concentration of dissolved organic matter, dissolved solids, and modulation of the microbiological population dynamics that can affect pathogen diversity, survival, and virulence (Zwolsman & van Bokhoven 2007; Thorne & Fenner 2008, 2011; Delpla et al. 2009; Emelko et al. 2011; Bogialli et al. 2013; Galway et al. 2015).

A changing climate is one of many drivers that affect the quality and availability of freshwater resources and can amplify, multiply, or temper threats imposed by concurrent other environmental stressors (National Research Council| (NRC) 2011; Intergovernmental Panel on Climate Change (IPCC) 2014). The capacity of conventional water treatment systems to respond or adapt to water quality perturbations depends on the inherent capacity of the treatment train to accommodate water quality changes that correspond to the intensity, frequency, and duration of each event in conjunction with watershed characteristics. From an adaptation perspective, it is important to understand the scope and magnitude of water quality changes that might affect the performance of drinking water treatment facilities. Quantification of performance impacts yields a basis to evaluate and, if necessary, increase the treatment capacity reserve in preparing for factors not originally considered in engineering. This paper explores the use of a generic model to assess the capacity of conventional water treatment processes (e.g., coagulation, sedimentation, and filtration) to adapt to climate-induced water quality changes. The types of location-specific data that are relevant for climate adaptation planning are illustrated using an example surface water treatment system.

Traditionally, design of water treatment facilities has been predicated on the underlying assumption that ample source water is available to meet all water use requirements (domestic, municipal, commercial, industrial, and agricultural) within the design life of the facility and that the quality and quantity of water available from a given source are relatively predictable based on historical data. However, increasing evidence of the lack of climate stationarity (Milly et al. 2008) is raising questions about the validity of traditional design assumptions, particularly since the service life of many facilities can exceed 50 years, typically well beyond the design life. Conventional treatment, which involves sequential coagulation, flocculation, sedimentation, filtration, and disinfection, has been widely used since the early part of the 20th century (Howe et al. 2012) and continues to dominate treatment practices used by the majority of surface water treatment facilities currently operating throughout the world. Variations among the individual treatment facilities include the type of pretreatment, chemical use, sedimentation designs, filtration designs, post-filtration treatment, and operating practices (e.g., backwashing strategies, solids management, monitoring, maintenance, etc.). Other technologies that are becoming more prevalent include advanced oxidation, adsorption, membrane technologies, ion exchange, biological treatment, ultraviolet irradiation, and/or alternative disinfection. However, conventional treatment with chlorine as a primary disinfectant is likely to remain the cornerstone of surface water treatment for the foreseeable future due to the high costs of upgrading or rebuilding existing treatment facilities.

Design parameters for conventional treatment systems include system flow rates, reactor dimensions and hydraulic properties such as mixing efficiency, hydraulic retention time, flow patterns, and filter media characteristics. Precipitation events and resulting changes in surface water quality can impact coagulation and clarifier performance (Hurst et al. 2004). The effectiveness of rapid sand filtration can be influenced by inconsistent and variable hydraulic loading rates and changes in the quality and quantity of solids (Glasgow & Wheatley 1999; Harrington et al. 2003).

Chemical types and dosage are a key component of conventional treatment trains and are used to accomplish multiple objectives such as coagulation, flocculation, precipitation, oxidation, or disinfection. Water treatment chemicals include metallic salts, organic polyelectrolytes, oxidants, and corrosion and scale inhibitors; however, it is important to recognize that there are limitations on the types of chemicals that can be used in drinking water applications. All chemicals must undergo testing to ensure compliance with health effects criteria (http://www.nsf.org/services/by-industry/water-wastewater/water-treatment-chemicals/nsf-ansi-standard-60/). Additional registration requirements have been implemented for disinfectants (e.g., antimicrobial chemicals) (http://www.epa.gov/oppad001/chemregindex.htm).

Typically, chemical selection and optimization is tailored to the source water characteristics. Final decisions are based on regulatory requirements, performance, availability of chemicals, costs, and the quantity of residuals that are generated. In general, chemical effectiveness and reliability are influenced by multiple interdependent factors, some of which are relevant to climate adaptation. A summary of operational and water quality factors that influence chemical effectiveness in conventional treatment systems is given in Table 1. The effectiveness of climate adaptation strategies depends on the ability to modify operations or chemical use in conjunction with episodic, intermittent, or continuous changes in source water quality.

Mathematical modeling approaches for evaluating water treatment facilities include simple calculations, spreadsheet-based models, computational fluid dynamics, and dynamic models that are coupled to the physical dimensions and process data of existing treatment units. The primary application of models is to systematize process analysis, evaluate the impacts of chemical addition on pH, removal of particles, and removal of disinfection byproduct (DBP) precursors (Hsu & Huang 2002; Chowdhury et al. 2009; Uzun et al. 2015). The capacity to model physical treatment components, such as mixing, sedimentation, and filtration, is fairly well established from mass balance calculations and reactor hydrodynamics (Rietveld & Dudley 2006). However, mathematical models of flocculation are dependent on data and are often difficult to generalize or extrapolate to other treatment facilities (Rietveld & Dudley 2006). Modeling of disinfectant chemistry, degradation rates, and formation of DBPs is also fairly well studied and supported by a wealth of empirical and compliance monitoring data (Ged et al. 2015). Another limitation of modeling approaches is that there is often a misalignment between modeling data requirements and the availability of monitoring data, which tend to be designed to support regulatory requirements and provide operational oversight.

From a climate adaptation perspective, specific changes in water quality and quantity depend on the interplay among the key drivers and their assimilative capacity within a watershed and climatic zone. Modeling frameworks can be used to explore the efficacy of some types of adaptation measures, such as modifying hydraulic loading rates, modifying mixing and recirculation parameters, introducing alternative chemical approaches, adding supplemental storage capacity, or other safeguards to protect infrastructure from impacts of meteorological events. The inputs required for modeling climate adaptation potential include historic, current, and projected water quality data and details on the treatment system. While there are uncertainties associated with climate projections, models can provide insights into the impacts of water quality variability on treatment effectiveness and identify vulnerabilities and priorities. There are several quantitative models available to assess climate change impacts on water treatment plants. In this paper, we apply two mechanistic models, the WTP-CCAM (Li et al. 2012, 2014) and EauSim (Rietveld & Dudley 2006), using data from a conventional surface water treatment facility. While each model contains inherent uncertainties, the underlying equations were used to evaluate the resilience of existing treatment systems to source water variability induced by climate, weather, or other causes. In addition, initial assessments of the extent to which the individual components of the treatment facility have the capacity to adapt to a changing climate were conducted.

To illustrate the concept of climate adaptation capacity, the Appalachian Plateau in the northeastern USA is used as an example watershed. A simplistic example of projected water quality changes in source water, shown in Table 2, provides a context for this analysis. In general, temperature, alkalinity, total organic carbon (TOC), and nutrient loading are expected to increase, while the hardness is expected to decrease (Li et al. 2012, 2014).

Data from a surface water treatment facility that is within this watershed, Greater Cincinnati Water Works Richard Miller Treatment Plant (GCWW) (http://www.cincinnati-oh.gov/water/about-greater-cincinnati-water-works/water-treatment/), were used to evaluate the efficacy of climate adaptation modeling. Available monitoring data were reviewed to identify a suite of source water quality data that could serve as proxies for climate-relevant data including alkalinity, hardness, total dissolved solids (TDS), turbidity, TOC, and pH. All data were supplied by the GCWW Miller Plant and had undergone internal quality assurance and data management protocols prior to being used in this case study.

Two models, WTP-CCAM (Li et al. 2012, 2014) and EauSim (Rietveld & Dudley 2006), were used to evaluate the consequences of five different water quality perturbations:

• temperature increase

• turbidity spike

• TOC spike

• alkalinity decrease

• TDS/ionic strength increase.

The responses of conventional treatment components (coagulation, flocculation, and sedimentation) were assessed using EauSim, while the downstream granular activated carbon (GAC) unit was analyzed using WTP-CCAM. The model inputs are summarized in Table 3 with information on the combination of treatment units that are relevant to the individual parameters. Process information (e.g., reactor dimensions, flowrates, reactor sequence) from the GCWW plant was used to model system performance. For the purposes of this analysis, a 5-year period was used to bracket the range of water quality and operational conditions. The extent to which the treatment system can accommodate deviations from these quasi-baseline conditions was modeled. Climate-relevant data can also be extrapolated from watersheds with similar land-use patterns.

The impacts of each of the five water quality perturbations modeled using the GCWW as an example system are summarized in Table 4 along with adaptation options. While this example is somewhat general, it provides a systematic approach to identify potential adaptation options and their range of effectiveness. For the GCWW water plant, the modeling and analysis described here show the potential to increase the treatment capacity of conventional processes. Other adaptation options for controlling specific disruptions such as changes in organic loading include upstream addition of powdered activated carbon (Carrière et al. 2009) or downstream use of granular activated carbon (Li et al. 2012, 2014). Modeling results can be used to streamline supplemental laboratory and pilot-testing of adaptation options.

While some general trends were observed through the modeled simulations, the results are limited by the lack of comprehensive chemical reaction information on the spectrum of chemicals that can be used for coagulation–flocculation and filtration aids. In addition, while the model provides a conservative estimate of the performance of sedimentation and filtration, it lacks the ability to fine-tune each treatment unit or the treatment system as a whole. Nevertheless, the use of models to evaluate impacts from different climate-related water quality changes provides an opportunity to identify climate adaptation options within an existing treatment infrastructure. Models also enable identification of limits beyond which the adaptation of existing infrastructure becomes ineffective. Such examples can be found in coastal areas with severe salt water intrusion and prone to coastal storm surges.

Adaptation of existing or new water supply systems to new climate realities is central to sustainability and resilience. Climate or other disruptive events can impact water quality in multiple concurrent ways. Additionally, accidental or intentional disruptions to water supplies can also compromise treatment performance. While the modeling example presented in this paper focused on individual parameters, the net resiliency depends on climate adaptation options that can span a range of water quality challenges. The capacity reserve concept, derived from engineering practice (Tillman et al. 1998, 2005; Dominguez & Gujer 2006; Matos et al. 2013), provides an organizing principle that could be useful for prioritizing climate adaptation strategies.

Water supply perturbations can result from multiple triggers. Adaptation options may range from major or minor modifications to existing treatment systems along with system-wide considerations such as off-line storage, operational changes in distribution systems, or the use of supplemental water sources including reclaimed or recycled water. The types of water supply challenges that can result from climate and weather-related events are summarized in Table 5. Climate adaptation options derived from the modeling framework are also summarized. In some cases, relatively available cost-effective solutions might be practicable, such as modifying chemical treatment approaches. In other cases, more comprehensive retrofits might be required.

Given the diverse nature of climate change impacts on surface water quality, conventional treatment processes are potentially economically viable adaptation candidates. To assure adequate capacity reserve, short-term climate adaptation options include chemical optimization, high-rate liquid–solid separation systems, and integration of biological filtration. Many of these adaptation options can be readily implemented with strategic modifications of chemical addition, mixing, and filter performance. While specific adaptation measures depend on watershed characteristics and treatment configurations, further analysis of several alternative approaches is warranted:

• targeted monitoring coupled with optimized chemical addition

• alternative coagulation/flocculation chemicals

• high rate liquid–solid separation systems

• integration of biological treatment.

More detailed characterization of waterborne particulates and organics is needed to optimize chemical selection and dosing. While bulk parameters such as turbidity and TOC are useful for routine monitoring, there is a need for improved capacity to model the impacts of weather and climate on these parameters. For example, the increased proliferation of cyanobacterial and algal blooms and associated toxins has been widely observed in surface water impoundments. While these organisms contribute to turbidity and TOC, their surface properties, chemical reactivity, and treatability differ from other types of turbidity and TOC. For example, increased turbidity loading may warrant modification of the coagulation/flocculation chemical regime (Kastl et al. 2004). Changes in particle size distributions may also affect coagulation/flocculation decisions. Decreased alkalinity may also necessitate tailored modification of chemical addition. Contaminants associated with runoff or sewer overflows may require additional pre- and post-treatment. Integrating upstream monitoring with treatment system models can lead to more resilient decision-support systems and more effective climate adaptation strategies.

The water industry relies on polyelectrolytes for multiple purposes with the dominant formulations consisting of polyacrylamides or polyDADMACs (Uyak & Toroz 2007; Matilainen et al. 2010; Padhye et al. 2011). However, newer formulations are emerging based on using nanotechnology to design coagulants that target specific functional groups. For example, polymers containing quinone or pyrrole side chains have been reported as effective scavengers of radicals and inhibiting nitrosamine production (Bao & Loeppky 1991; Wang et al. 1994; Zeng et al. 2014). There has also been increasing interest in polyelectrolytes derived from natural products such as histidine, chitosan, and soy protein. These types of tailored polymers have the potential to substitute for polyDADMACs or other polyelectrolytes; however, it is important to ensure that alternative formulations meet National Sanitation Foundation (NSF) requirements and do not lead to other inadvertent reactions or DBP formation during treatment, storage, or distribution. Coagulation effectiveness can also be enhanced with strategic use of coagulants (Ma & Liu 2002; Jiang & Wang 2003) There are also concerns about DBP formation from alternative coagulants such as NDMA (Kinicannon et al. 2003).

There have been significant advances in high-rate sedimentation systems that can be retrofitted into existing facilities. For example, ballasted sedimentation, dissolved air flotation, or plate separators optimize sedimentation through increasing the settling velocity (ballasted sedimentation), decreasing the particle density (dissolved air flotation), or decreasing the rate of particle removal (plate separators). Integrated systems, such as the use of magnetically enhanced ion exchange, have been demonstrated to remove DBP precursors in conjunction with particle removal (Watson et al. 2015). There is a need to integrate these types of technologies into modeling frameworks to evaluate their potential roles in climate adaptation, costs, and capacity to respond to changing water quality.

Biological treatment systems have a long history in controlling biodegradable organics in wastewater and stormwater. Biological systems have also been used indirectly in water treatment as a result of biofilm development on surfaces and microbial colonization of filter media, adsorbents, and ion exchange resins. The augmentation of conventional treatment with microbial processes holds promise for climate adaptation. Microbial processes can be effective at mineralizing organics, nitrification and denitrification, and mediating oxidation-reduction reactions for removal of endocrine active compounds and other trace contaminants. From the perspective of climate adaptation, the elevated temperatures and organic content that are associated with a changing climate are well suited for biologically enhanced treatment systems. However, it is important that biological treatment systems do not harbor pathogenic organisms that could impair water quality. In addition, the microbial biomass can serve as precursor material for DBP formation including NDMA. Models and design data for biological filtration systems are continuing to develop.

Scenario I reflects an acute and temporary event such as a sewer overflow or chemical spill (Whelton et al. 2015) where there may be a lapse in the capacity to provide safe drinking water. Scenario II represents a lack of resilience to extreme weather events and their aftermaths, such as hurricanes, intense storms, flooding, or drought conditions (Pardue et al. 2005; Brozovic et al. 2007; Yoon & Raymond 2012; Dhillon & Inamdar 2013). Scenarios III and IV reflect long-term climate change impacts on source water quality that may result in a ‘tipping point’ being reached due to inadequate capacity to implement operational changes (increasing chemical feeds, extended pumping, additional finished water storage, etc.) or treatment plant upgrade. Recovery from Scenario IV may necessitate significant capital improvements such as additional treatment unit processes, source water storage, distributed network modifications, or paradigm shifting (direct/indirect reuse) to avoid future recurrence of the service disruptions.

Conventional water treatment processes (e.g., coagulation, sedimentation, and filtration) have a long history in producing drinking water from surface water supplies. As water quality changes in response to a changing climate, the adaptation capacity of the existing infrastructure should be considered. Case-specific analysis and process simulation may provide insightful information to develop engineering options for climate adaptation that incorporate watershed-specific conditions. Even though many models are based on steady-state assumptions and lack a direct way to accommodate water quality fluctuations, basic insights can be gained into capacity reserve. Harmonizing of monitoring data with modeling parameters is important for calibrating and validating models. Further insights can be gained through integrating upstream conditions. In addition, more robust data and models are needed to investigate the spectrum of chemical and microbiological reactions that are associated with the complete inventory of chemical amendments that can be used in water treatment practice. Integration of biological treatment models is also an important component of evaluating climate adaptation options and capacity reserve.

This research was supported through the US Environmental Protection Agency (EPA) Water Resources Adaptation Program and the US National Science Foundation (NSF) Independent Research and Development Program. The authors are solely responsible for the content and writing of this paper, and any opinions, findings, conclusions, or recommendations are those of the authors and do not necessarily reflect the views or policies of the US EPA or the NSF.

The US Environmental Protection Agency (EPA) through its Office of Research and Development funded and managed the research described herein under EPA Contract EP-C-11-006 to Pegasus Technical Services. It has been reviewed by the Agency but does not necessarily reflect the Agency's views. No official endorsement should be inferred. EPA does not endorse the purchase or sale of any commercial products or services.