Energy and water quality management systems for water utility's operations: a review.

Journal of Environmental Management 153 (2015) 108e120

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Review

Energy and water quality management systems for water utility's operations: A review Carla Cherchi a, *, Mohammad Badruzzaman a, Joan Oppenheimer a, Christopher M. Bros b, Joseph G. Jacangelo a, c a b c

MWH Americas, Inc., 300 N. Lake Ave, Suite 400, Pasadena, CA 91101, USA MWH Global, Inc., Temple Court, Birchwood, Warrington WA3 6GD, UK The Johns Hopkins University Bloomberg School of Public Health, 615 N. Wolfe Street, Baltimore, MD 21205, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 31 July 2014 Received in revised form 13 January 2015 Accepted 31 January 2015 Available online

Holistic management of water and energy resources is critical for water utilities facing increasing energy prices, water supply shortage and stringent regulatory requirements. In the early 1990s, the concept of an integrated Energy and Water Quality Management System (EWQMS) was developed as an operational optimization framework for solving water quality, water supply and energy management problems simultaneously. Approximately twenty water utilities have implemented an EWQMS by interfacing commercial or inehouse software optimization programs with existing control systems. For utilities with an installed EWQMS, operating cost savings of 8e15% have been reported due to higher use of cheaper tariff periods and better operating efficiencies, resulting in the reduction in energy consumption of ~6 e9%. This review provides the current state-of-knowledge on EWQMS typical structural features and operational strategies and benefits and drawbacks are analyzed. The review also highlights the challenges encountered during installation and implementation of EWQMS and identifies the knowledge gaps that should motivate new research efforts. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Energy and water quality management system Water utilities Energy efficiency Water quality Demand forecasting

1. Introduction The water and energy sectors are being inextricably linked through a co-dependent and complex relation of mutual exchange of resources often referred as the Water-Energy Nexus (Raucher et al., 2008). High volumes of water are required to generate energy and conversely, water abstraction, treatment and distribution are highly energy-intensive processes [Rothausen and Conway, 2011; Griffiths-Sattenspiel and Wilson, 2009]. Water and energy are critical resources, and their integrated management can provide important economic and environmental benefits in both sectors [Wilkinson, 2008; Kanakoudis et al., 2012]. Over the past several decades, water organizations have been challenged by new stringent regulatory requirements, increasing energy costs and demands, and decreased availability of high quality source water [Sovacool and Sovacool, 2009]. As freshwater becomes scarce, more energy is required to extract water from

* Corresponding author. E-mail address: [email protected] (C. Cherchi). http://dx.doi.org/10.1016/j.jenvman.2015.01.051 0301-4797/© 2015 Elsevier Ltd. All rights reserved.

aquifers, process saline water into potable drinking water and deliver freshwater over long distances [Wilkinson, 2008]. Water utilities have become increasingly energy intensive and responsible for an approximate 3% share of U.S. annual electricity consumption, which increase to as high as 13% when residential water use is included [Boulos and Bros, 2010; U.S. EPA, 2012; Sanders and Webber, 2012]. Future projections estimate this percentage to double to 6% due to higher water demand and more energy intensive treatment processes [Chaudhry and Shrier, 2010]. Estimates indicate that approximately 90% of the electricity purchased by U.S. water utilities, US$10 billion per year, is required for pumping water through the various stages of extraction, treatment, and final distribution to consumers [Bunn, 2011; Skeens et al., 2009]. Further, the energy use in water utilities, with the exclusion of energy use for water heating by residential and commercial users, contributes significantly to an increasing carbon footprint with an estimated 45 million tons of greenhouse gas (GHGs) emitted annually in U.S. into the atmosphere [Griffiths-Sattenspiel and Wilson, 2009; Wallis et al., 2008; Kanakoudis, 2014]. In anticipation of federal and state legislation that may impact future GHG emissions from water facilities, it is important that

C. Cherchi et al. / Journal of Environmental Management 153 (2015) 108e120

energy management practices become an integral part of effective water utility management. Current management strategies, often, pose complex coordination issues between energy savings, water quality concerns, and operational and maintenance issues [Barry, 2007; Jentgen et al., 2003]. For water utilities, source supply and distribution systems control and operations have typically followed consumption [Jentgen et al., 2003] and have mostly aimed to solely achieve water quality goals with less attention oriented towards energy costs and carbon footprint reduction [Bunn and Hillebrand, 2008; Kanakoudis and Gonelas, 2014]. Recently, smart water grid concepts have been developed to enable better management of the water network by leakage and pressure management, capital spending optimization, streamlined water quality monitoring, and network operations and maintenance, however without addressing energy management issues. The concept of an integrated Energy and Water Quality Management System (EWQMS) was introduced in the early ‘90s to provide water utilities with the foundation for a systems control management tool for simultaneously achieving energy efficiency and water quality objectives [Jentgen et al., 2003]. An EWQMS is a collection of individual application software programs, useredeveloped or commercially available, that allow the implementation of an array of energy cost reduction strategies operating within designated constraints. Real‒time communications with preeexisting SCADA (Supervisory Control and Data Acquisition) systems allow the EWQMS software to monitor and proactively provide recommendations regarding system operation (e.g., pumping, storage tank turnover, etc.) based on timeeofeday electrical use and associated tariff, forecasted demand and pump scheduling [Barnett et al., 2004]. Two decades after the EWQMS concept was first introduced, approximately twenty drinking water utilities across the world (US, New Zealand, Canada, Australia, South Korea) have installed the EWQMS architecture at their facilities, confirming an increasing interest in EWQMS as energy prices and demand continue to increase [Badruzzaman et al., 2014]. Although design and implementation of this platform are characterized by numerous challenges, EWQMS provides a number of economic, environmental and operational benefits, such in more effective risk management for maintenance of water quality objectives and favorable costebenefit solutions. Utilities with installed EWQMS have opportunities to better manage their water supply portfolios and their local water resources. EWQMS can be employed as a system simulator for decision making that assists engineers and planners in understanding the impact of water demand patterns and energy market profiles on their water resources management. EWQMS also assists water utilities in monitoring the water balance, identifying water losses in the distribution systems, thus reducing overall water consumption. Water leakage from aging infrastructure poses a challenge for water supplies, particularly in areas that are struggling to support the growing demand [Kanakoudis et al., 2013a; Kanakoudis et al., 2014]. The daily water loss through leakage, which is estimated to represent about 40%e 50% of the total daily water consumption [Tucciarelli et al., 1999], is associated to the waste of a large amount of energy [Nasirian et al., 2013]. To date, there is no comprehensive review paper on EWQMS in the peerereviewed literature; there exists only sparse and fragmented documentation of EWQMS practices within the water industry. Because this field of study and implementation is growing rapidly, a review of the various aspects of EWQMS is warranted. Thus, this review aims to collectively integrate the fundamental concepts of energy, water demand and supply, and water resources management to provide a more perspicuous understanding of the structure of the EWQMS platform at drinking water utilities. The

109

specific objectives of this review are to: Discuss the historical development and modules of EWQMS; Assess the major optimization strategies for EWQMS operation; Illustrate typical mathematical models used for the system optimization; Identify the benefits of EWQMS installation at selected water utilities; Analyze the challenges encountered by water utilities during EWQMS implementation and operation; and Identify the knowledge gaps and future research needs for more effective EWQMS practices.

2. Historical development and structure of EWQMS 2.1. Historical development of EWQMS Pump energy management systems have been investigated by the water industry and academic researchers for more than two decades. In the early 1990s, the AWWA Research Foundation, Electric Power Research Institute's Community Environmental Center and the East Bay Municipal Utility District (California, US) (EBMUD) funded the first EWQMS project in the U.S. [Morley et al., 2009] and outlined the key components of a practical EWQMS for water facilities. In 1996, a group of water and electric utilities, academics and consulting engineers began developing the functional specifications of a more formalized EWQMS prototype followed by the software installation at EBMUD. However, the difficulties encountered during the implementation required additional research efforts. In 2003, Colorado Springs Utilities developed an offeline EWQMS by building on the lessons learned from the EBMUD prototype, creating a customized Operations Planning Scheduler and related organizational processes [Jentgen et al., 2003]. This offeline study demonstrated the feasibility of an EWQMS for control of the system's daily operations. In the years 2000e2004, the installation of commercially developed software at Wellington (New Zealand) and EBMUD were the first of several EWQMS implementations at other water utilities across the world (e.g., UK, Canada, Australia, Korea), as shown in Fig. 1. Both commercial software and researchegrade optimization techniques from various academic efforts have been developed over the years. Finesse [Rance et al., 2001], POWADIMA [Salomons et al., 2007], and Neptune [Morley et al., 2009] are software applications principally developed with government (e.g., European Union) and water industry funding. These efforts focused on developing suitable algorithms for management of real, complex water supply systems by undertaking proof of concept studies using either off-line post processing or limited on-line real time control. Few academic software applications have been used for full-scale operation of major supply systems. However, commercial software suppliers, such as Derceto's Aquadapt, have achieved some success with real, extended operational examples [Derceto, 2013]. In addition to Derceto's Aquadapt, Schneider Electric's Aquis, Tynemarch's MISERePSL [Fowler and Main, 2010; Woodward and Fowler, 2011] and Innovyze's IWLive [Innovyze, Inc., 2013] are other commercial software products that are marketed as EWQMS packages that typically interface with commercial hydraulic models. These products use different programming options to optimize pump scheduling. While Project Neptune pump optimization is formulated on Model Predictive Control, MISERPSL, Aquadapt and Aquis employ linear, mixed integer linear and dynamic programming respectively to control pump operations (Skworcow et al., 2009; Fowler and Main, 2010; Derceto, 2013). BalanceNet, an add-on module of IWLive, overcomes the problem

110


Fig. 1. Timeline of historical development and implementation of EWQMS in water utilities worldwide.

of extensive search space faced by metaheuristic solvers by performing a two-stage optimization that includes a linear mass balance followed by a pump selection based on a genetic algorithm. Although most of the utilities are currently utilizing commercial software platforms for their EWQMS (e.g., Aquadapt from Derceto, Inc.), a number of water utilities such as Las Vegas Valley Water District [Murphy et al., 2007], Yorkshire Water Services [Penny, 2011], Albuquerque Bernalillo County Water Utility [EMA, 2013], Metro Vancouver [EMA, 2013] and JEA [Jentgen et al., 2005] have developed solutions inehouse.

2.2. Structure of an EWQMS An EWQMS is considered a collection of individual application software programs that interface with existing monitoring and control systems to cohesively address water quality, supply and energy management within operational constraints [Jentgen et al., 2003]. A typical framework consists of similar structural modules regardless of differences in operational characteristics at different utility installations. The schematic in Fig. 2 illustrates the interconnection of typical EWQMS components in a conceptual framework and those that are typically integrated in commercial software products and an in-house (LVVWD) architecture. Briefly, water utility operations are coordinated by an Operation Planner and Scheduler whose primary role is to develop a daily operating plan for the entire system [Jentgen et al., 2003]. The software integrates and communicates with the SCADA and the approach

relies on hydraulic predictions and accurate forecasting of hydraulic performance, pump performance, electricity tariff structure changes, system demand and water quality [Jentgen et al., 2003]. 3. Fundamental optimization strategies of an EWQMS The primary goal of an energy and water quality management strategy has been to minimize energy costs without compromising on the water quality. Recently, the energy price escalation and volatility has created budgeting difficulties for utilities [NYSERDA, 2010]. Hence, to reduce energy bills, water utilities have engaged in a broad array of programs based on energy efficient practices, the use of alternative and cleaner energy sources, and/or onesite generation [Bunn et al., 2006]. In particular, the shift of the electrical load (e.g., for pumping) to lower cost tariff periods has been advantageous for those utilities that are subject to timeeofeuse (TOU) charges by electricity providers [Raucher et al., 2008]. An EWQMS provides a framework to implement water utility's energy management strategies on a real‒time basis, which to date has been achieved with optimization steps taking as little as thirty minutes. Further details on energy cost reduction opportunities that have been considered in EWQMS operations are discussed below. 3.1. Optimizing operation around tariff structures Electric utility rates have been increasing over time [Brueck and


111

Fig. 2. Conceptual schematic of a EWQMS module [Adapted from Jentgen et al., 2003 and Derceto, 2013]. The gray shaded area indicates typical components of a commercial software EWQMS tool. Red dashed area refers to the in-house LVVWD software architecture. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Geisenhoff, 2010]. Understanding electric utility rate structures is essential for efficient energy cost management of complex water operations [NYSERDA, 2010]. In general, 30e60% of the electric cost is based upon the demand for electricity, with the remaining cost based on the actual energy used [NYSERDA, 2010]. Different energy markets offer water utilities a variety of electricity purchasing options, which may consist of: i) standard contracts with variable or flat annual rates; ii) timeeofeuse (electricity generation price varies depending on the season, day and time-of-day); iii) energy and demand charges (billing includes charges for the amount of electricity consumed and respective rate of consumption during the billing period); iv) criticalepeak pricing (electricity prices may reflect the costs of generating and/or purchasing electricity at the wholesale level); v) demand bidding (load reduction bids on an hourly basis for specific events without financial penalties); vi) negotiated hedge contracts for fixed blocks of energy use; and vii) and real‒time electricity market options such as hourly prices set in a dayeahead market [RMI, 2006]. In some markets, these different options may be applied to specific portions of a distribution system, i.e., one part may be running on a flat rate tariff while another is operating on a TOU contract. By reviewing the electricity rates available for a given service area, the EWQMS identifies and solves for the most economical combination to minimize energy costs. Many water utilities that have implemented EWQMS are

charged with TOU options that offer considerably lower rates during “offepeak” periods when the demand is lower, and reward customers that have predictable energy demands over time [Thorstensen, 2007; NYSERDA, 2010]. An effective EWQMS system responds to TOU charges by shifting major electrical demand to maximize the use of lower electricity pricing within water quality and operational constraints [Bunn et al., 2006). Electrical loads are often moved to lower cost tariff blocks (e.g., overnight), for intraeday operations, or from season to season where longeterm raw water storage is possible [Raucher et al., 2008]. This practice is most available to water providers with sufficient storage to shift pumping loads to offepeak hours and then drawing from storage tanks or reservoirs during high cost periods [Bunn and Reynolds, 2009; Raucher et al., 2008]. Water utilities can also significantly reduce their electricity bills by flattening the energy demand curve, particularly during peak pricing periods [NYSERDA, 2010]. For water utilities that use natural gas, gas tariffs contribute to the utility's annual bills. In California, natural gas does not usually have TOU components and, during summer, its purchase price is generally lower than that of electricity [Water Energy Innovation, 2013]. Additionally, natural gas charges are less volatile over long periods of time and they are not subjected to realetime supply and unpredicted demand influences [Water Energy Innovation, 2013]. Recently, Demand Response (DR) Programs have been

112


developed by electric utilities to promote an efficient distribution of electricity to endeusers while providing incentives for demand reduction [PLMA, 2002] and water utilities are increasingly cooperating with the electricity supplier's DR programs to manage their energy consumption. In addition, DR programs balance supply and demand in realetime, which help overcome the uncertainties of intermittent electricity generation, particularly when provided by renewable sources, such as wind energy, that are unpredictable €rkka €inen and Ika €heimo, 2009]. EWQMS evaluates all tariff rate [Ka structures including DR programs (if applicable) and finds the lowest energy cost solution. 3.2. Selecting low cost energy mix Water utilities often diversify their energy portfolios not only to increase their electricity reliability, but also to increase their energy efficiency and reduce their contribution to GHG emissions [Raucher et al., 2008]. Among the various opportunities available, ineconduit hydropower generators, natural gasedriven pumps and alternative renewable options on-site and connected to the grid, such as solar and wind, are the most utilized by water utilities [Klein, 2009]. The EWQMS optimization finds the lowest cost energy mix within the operational constraints. Hydropower is one of the least expensive energy generating options that water utilities are considering, as it plays an important role in stabilizing the electrical transmission grids and in meeting peak loads, reserve requirements and additional ancillary needs [Jentgen et al., 2003]. New “microhydro” applications (low head dams or hydrodynamic propellers) have also been introduced on existing water storage or conveyance structures and typically provide up to 100 kW of electricity using natural water flows without the purchase of fuel [Chaudhry and Shrier, 2010]. If available, EWQMS strategies can be used to maximize the value of hydroelectric generation to gain financial benefits. Colorado Springs Utilities (Colorado Springs, U.S.) is an example of how a prototype EWQMS could reach substantial dollar savings by maximizing the supply of electricity from the 28 MW Tesla hydroelectric plant [Jentgen et al., 2003]. During peak demand periods (e.g., summer), water utilities can reduce their electrical demand, thus seasonal electric costs and risks, by operating natural gasedriven power in a sole or a dual fuel mode. Natural gas is one of the major fuels consumed by the water sector within the U.S. (17 Million Therms in 2010, Southern California Gas Company estimation) predominantly to power pumping [Water Energy Innovation, 2013]. Gas engines are remarkably less efficient than electric motors (18e30% compared with 95%); however, they may provide approximately 52% cost savings over electricity [Goonetilleke, 2011]. Thus, the natural gas alternative only becomes more economically acceptable if it is less costly than electricity in delivering the same volume of water [Bunn et al., 2006]; such an analysis is usually determined through common breakpoint analysis approaches. Eastern Municipal Water District (EMWD) in California, U.S., uses their EWQMS to balance a mix of electric and gas driven power (e.g., pumping) to avoid peak demand electricity charges. In addition to natural gas, diesel is often considered as an electricity load shaving alternative and is a common backup generation option for water utilities in noneattainment zones under the Clean Air Act. Water facilities with renewable energy assets and under EWQMS automation have opted for shifting power demand to the available renewable source (e.g., wind) at night and favoring solar contributions during daytime hours [Chaudhry and Shrier, 2010]. The renewable source selection and time of use is based on a realetime or programmed interval comparison of the price of fuel, the price of purchased electricity and the price of netemetered

electricity sold back to the grid. California water utilities, for example, have been able to sell renewable energy back to their electricity provider during peak times and purchase electricity during offepeak periods. 3.3. Selecting operating schemes to lower energy consumption Energy cost reduction by improving energy efficiency of operation is one of the EWQMS objectives. In most EWQMS operations, efficient pump scheduling and network optimization are significant contributors to efficiency practices. 3.3.1. Energy efficient operation of pump stations An EWQMS reduces energy costs by operating pumping systems more efficiently than simple level or pressure controls. This approach is particularly important in energy tariff markets where the differential between peak and offepeak rates is low. In such cases, the lowest cost strategy is associated with running pumps efficiently in peak periods rather than running inefficiently in offepeak periods [Reynolds and Bunn, 2010]. In order to ensure energy efficiency of a pump station, a comprehensive understanding of the system requirements (e.g., minimum, average, and peakeday demand; neareterm and future demand; water age and duration of operation) are important. Unknowns such as a change in demographics, deterioration of pump performance, pipeline fouling and scaling, and future water demands are difficult to anticipate and tend to create conservatism in pump design [Kanakoudis and Gonelas, 2014]. Conservative design assumptions (e.g., maximum flow and worstecase pipe friction factors) may result in oversized impellers, oversized motors, and pumps that may operate away from their best efficiency point (BEP). It is important to tailor the pumps to the system operating requirements to ensure that the BEP is located close to design or near the operating point of the system curve where the pump operates most often [Reynolds and Bunn, 2010; Senon et al., 2013]. The energy efficiency of a pumping system might be restored or improved by pump refurbishments, operational changes, pump replacement, or a closer match with the duty point. One of the benefits associated with the implementation of an EWQMS is the capability to select and operate pumps close to their BEP at minimum to maximum flow conditions on a realetime basis. Historically, pumps have been scheduled based on the “maximum flow” (i.e., run the pump until it can no longer handle the system requirement) or based on the “percent of maximum speed” (i.e., select a percent of the maximum speed and select the combination of pumps based on that criteria). However, none of these traditional methods provide an accurate estimation of the operational efficiency of the pumping system. In applications where multiple pumps are arranged in parallel operating in a leadelag sequence, the specific energy (energy consumed per unit volume of water pumped as expressed in kWh/MG) is used in the EWQMS to determine the most energy efficient control timing to start or stop a lag pump. Therefore, the combination of pumps is selected in a manner that minimizes the specific energy consumption of the entire station within operational constraints. A commercial EWQMS tool has the ability to store calibrated pump and efficiency curves in order to predict pump operational performance relative to pump curves for half hour intervals over a 24 h period [Reynolds and Bunn, 2010]. Selection of pumps based on the lowest specific energy consumption is still prone to a number of constraints in any real world system. These include: minimum run times; maximum starts per hour; minimum cool down times; minimum flow rates and maximum discharge pressures for valve stations; minimum and maximum plant production rates; pump station pressure rules (e.g., starting the smallest pumps first); and pump stagger timing to


prevent surges or water hammering. 3.3.2. Network optimization A significant percentage of energy input to a water distribution system is lost in pipes due to friction, pressure and flow controls valves, and consumer taps [Innovyze, 2013]. Network modifications and changes in operating rules may be combined with water resource and treatment management planning to provide a holistic sourceetoetap assessment. Thus, EWQMS is best implemented in parallel with conventional energy audits, pump replacements or refurbishment and the analysis of system constraints and pinch points. In general, water network optimization practices vary according to utility needs. Table 1 shows typical network and operational modifications in water utility optimization practices. Changes made for energy management purposes also need to comply with water quality, flexibility and security objectives; thus, risk management plays a role in ensuring a balance between energy optimization and operational flexibility. Various metrics can be used to assess Water Network Energy Efficiency, including the evaluation of kWh/MG supplied (or gCO2/ MG), the proportion of input energy utilized in pumping (wireetoewater efficiency) and transport (friction, tap and discrete energy losses) and, more broadly, the comparison of the minimum theoretical energy required with actual system performance ndez et al., 2010; Boulos and Bros, 2010; Kanakoudis et al., [Herna 2013b]. An understanding of the energy input and the spatial utilization of energy in water supply systems is required for water network optimization and accounting of associated energy savings. In general, only the operating areas with the greatest potential for savings are subject to optimization. The energy distribution in a system can be estimated using hydraulic network models, such as EPANET [Rossman, 1994] or embedded energy calculations (e.g., Water Network Energy Efficiency concept) in existing commercial software (such as InfoWater and InfoWorks WS, Innovyze). Most commercial EWQMS software does not have builtein hydraulic network modeling. In other cases, a hydraulic model is integrated with the EWQMS system to identify energy efficient routes for water transport through the distribution system. For instance, Aquadapt finds the shortest path to supply water to the distribution system network, reducing the number of times the same water is pumped to reach its final destination Table 1 Typical network and operational modifications employed in water utility optimization [Raucher et al., 2008; Leiby and Burke, 2011]. Network modifications

Operational modifications

Reconfiguration of pressure zones (e.g., revalving or installation of new control valves, changes to force main connectivities); Removal of hydraulic bottlenecks or pinch points (e.g., main reinforcement, replacement or new connections); Pump and motor refurbishment or replacement (e.g., use of variable frequency drives,VFDs); Ineline booster pumping stations with or without disinfectant; Construction of new or extension of existing reservoirs or hydropneumatic tanks (e.g., improve circulation); Use of Energy Recovery Units (ERUs) (e.g., in-line turbines) to generate grid power from the water supply network.

Water resource management (e.g., exploiting low cost sources); Water treatment optimization (e.g., improving chemical usage and in eprocess power use); Pump scheduling (e.g., improved manual or local closedeloop operating rules); Storage tank cycling (e.g., reducing retention times); Water quality management (e.g., flushing schedules or dosing optimization); Leakage management (e.g., active leakage control); Pressure management.

113

[Thorstensen, 2007]. In a welleconnected network system where water can be easily moved through multiple transport routes, the shortest flow path might provide a more energy efficient operation. 3.4. Water quality and water supply management strategies The operation of an EWQMS assists in protecting public health and safety by managing water utility operational risks to water quality [Bunn, 2006], while simultaneously reducing energy consumption and production costs. Drinking water practices that are based on energy efficient programs must still comply with federal and state environmental quality standards and achieve plant production requirements [Raucher et al., 2008]. Thus, customized strategies that are part of a “multiple barrier” approach [Raucher et al., 2008] have to be integrated with the EWQMS in order to select and manage raw water supply sources and maintain statutory water quality levels within the distribution infrastructure. 3.4.1. Source water quality and supply Water supply management objectives for operational cost minimization must also ensure that source water characteristics will not have significant impacts on water quality before final delivery to customers [Jentgen et al., 2003]. Water sources are expected to fully satisfy the distribution demand profiles by feeding a pressurized distribution network. Selection of raw water sources by EWQMS is contingent upon water rights and on the agreement that each district stipulates with water providers for provisions of the selected source portfolios. The implementation of a Water Supply Analyzer within the EWQMS can substantially improve the daily management of different water sources that supply the water system for a specific utility [Jentgen et al., 2003]. A better management of water sources will not only result in lower treatment costs, but will enable water utilities to better utilize limited resources and to provide optimal service to their customers. The availability and allocation of water from these sources are dependent upon seasonal and climatic conditions, raw water chemistry and quality and technologies available for treatment (e.g., in desalinated sources) [Kanakoudis, 2004]. This input leads the EWQMS optimization software towards the selection of water quality supplies within the constraints of water quality and costs. Appropriate pump operation for raw water withdrawal from selected sources is scheduled to reduce costs and improve the level of water quality in downstream processes. Optimal management of sources and reservoirs is expected to yield higher water quality, particularly when it supports a water treatment plant production with a constant flow rate [Bakker et al., 2011]. Main water quality improvements and reduced overall energy consumption are observed when the automation of reservoir control is based on model prediction rather than level based flow controls [Bakker et al., 2011]. This guarantees minimal changes in production flow and up to 10e20% lower turbidity and particle counts by effectively setting upper and lower reservoir levels under the flow forecasted conditions. The quality of raw water sources may influence the treatment processes required. Although water treatment processes are highly capital intensive components of the drinking water cycle, their optimization has not been integrated with EWQMS programs since treatment energy expenditure is negligible compared to pumping and transmission. 3.4.2. Distribution system water quality Distribution system operation and management must comply with specific criteria to avoid water quality degradation and loss of reliability during transmission. Low flow velocities in distribution systems may lead to water stagnation and failure of system flushing

114


mechanisms. High retention times due to large storage volumes will increase the risk of decreased turnover, with accompanying disinfection by-product (DBP) formation and biological reegrowth. Thus, theoretical models consider factors such as residual disinfectant concentration, DBP formation potential, and taste and odors for distribution system optimization [Kurek and Ostfeld, 2013]. Modeling water quality based on chlorine residuals and DBPs is not common practice due to monitoring constraints; water age is typically used as an indicator parameter of water quality and recommended as a key decision tool for water distribution system management [Kurek and Ostfeld, 2013]. Modeling water age is relatively simple, since its calibration does not require additional parameters apart from those required for hydraulic calibration [Murphy et al., 2007; Soyupak et al., 2011; Patki et al., 2013]. However, water age may not accurately reflect chlorine residual and DBP profiles if variability of chlorine decay rates exists in different segments of the distribution system, if multiple water sources are withdrawn and if waters of different ages are combined in tanks or pipe junctions [Murphy et al., 2007]. If not adequately managed, typical energy efficiency practices can increase or decrease water age in the system, particularly when storage tank refills are shifted to offepeak hours, when booster pumps are not adequately set to guarantee sufficient pressure [Raucher et al., 2008] or if multiepressure zones are not managed with properly configured pressure regulating valves. Storage issues, due to tank short circuiting and dead zones, can sometimes become too complex for hydraulic modeling software to simulate water age during normal extended period simulation runs, thereby adding to the computational requirements. In some instances, distributed processor simulations may overcome this limitation. Current EWQMS programs incorporate solutions that avoid increasing water age by utilizing synchronized pump withdrawals from reservoirs of the same cascade or by decreasing reservoir volumes by lowering the high and low operating levels, thus increasing the percentage of reservoir volume occupied by fresh incoming water [Beyer et al., 2005]. Schedules that follow deep cycling and pulse flow operation are also considered when reservoir storage is insufficient to cover the following 24 h needs of a specific pressure zone [Beyer et al., 2005]. 4. Objective function formulation and optimization techniques 4.1. Tradeoffs between energy, water quality and operations management In EWQMS, best efficiency practices for energy and cost

reduction are simultaneously implemented within the boundaries of water quality requirements and operating rules. Table 2 shows a conceptual illustration of the effect of various energy management strategies on energy cost, energy consumption, GHG emissions, water quality, and overall system reliability. The table indicates that the implementation of energy management practices that target the reduction of costs may positively or negatively impact the other operational improvement goals. Because of the interdependency and conflicting relationships between objectives and risk management practices, the optimization is critical for effective operation. Prediction and system optimization in an EWQMS are attained using an array of algorithms that solve optimization problems for specific objective functions (e.g., goals) by imposing distinct constraints [Boulos et al., 2006]. The use of complex algorithms within an extensive number of operational constraints may not guarantee an optimal solution and are prone to saturate hardware and software capabilities by generating timeeintensive computations [Bunn, 2006; Raucher et al., 2008]. The cost of energy, rather than energy consumption, is the objective function typically minimized for utilities in order to achieve lower operating costs. Fig. 3 shows a conceptual illustration of some of the constraints that are considered as boundary conditions while generating a solution for a least energy cost operation. These constraints ensure compliance with water quality regulations, consumptive use criteria and energy and mass conservation principles in the network system and at junction nodes [Boulos et al., 2006]. Energy charges, energy efficient measures and asset management practices are other implicit and explicit boundaries typically applied. Bunn [2006] discussed the major operational constraints of four U.S. water utilities by emphasizing the distinctive features of each optimized scenario. Some examples include deep cycling of water tanks to reduce average water age (e.g., EBMUD), the use of gas reciprocating engine driven variable speed pumps (e.g., EMWD), and reservoir management to maximize limited storage system capacity (e.g., Washington Suburban Sanitary Commission, WaterOne). 4.2. Optimization techniques for EWQMS A number of mathematical methods have been successfully applied in water management to solve optimization problems [Bagirov et al., 2012]. Deterministic techniques (e.g., linear, nonelinear and dynamic programming) have commonly been applied due to their high robustness and low computational requirements [Jowitt, 1992; Nace, 2001]. In linear programming, linear functions are used to describe the problem constraints and the objective functions, such as those used for minimizing satellite booster chlorination and pumping to a fixed free outfall [Boulos et al.,

Table 2 Effect of energy and water quality management strategies on overall water utility management objectives [Adapted from Raucher et al., 2008].

Shift pumping from peak to low cost tariff periods Select natural gas driven engines during peak tariff periods Utilize renewable energy generation onesite Optimize pumping strategies and system operations (e.g., pump combination selection, VFD installation, pumping in cascade, pumping at/near the BEP, etc.) Reduce leakage in distribution systems (e.g., pressure reduction strategies and infrastructure improvements, etc.) Select high quality source water Manage reservoir or storage tank (e.g., new setepoints for reservoir's operating levels, new water withdrawal management strategies such as deep cycling and pulse flow operation, etc.)

Energy cost

Energy consumption

GHG emissions

Water quality

System reliability

Decrease Decrease

No Change Increase

NQ NQ

Worsen No Change

Worsen Improve

Increase Decrease

No Change Decrease

Decrease Decrease

No Change Improve

Improve Improve

Decrease

Decrease

Decrease

No Change

Improve

NQ Decrease

NQ NQ

NQ NQ

Improve NQ

NQ Improve

Note: “NQ” (Not Quantified) indicates that the impact has not been quantified or that a positive or negative impact may occur depending on site specific constraints.


115

Fig. 3. Objective functions and typical operational constraints of an EWQMS. *Currently not included as an operational objective function or a constraint in real EWQMS implementations

2006]. However, the nonelinear nature of the majority of optimization problems in drinking water practice limits the effective applicability of these linear approaches to nonelinear systems of high complexity, particularly if based on nonelinear hydraulic and water quality relationships. Typical EWQMS systems based on linear and nonelinear approaches employ mixed integer programming to solve pump scheduling problems that are based on binary choices (e.g., on/off) [Derceto, 2013; Bagirov et al., 2012]. In dynamic programming, the system is discretized into a series of stages within an allowable range and generates a discrete solution that may not correspond to the global solution of the initial problem. Dynamic programming is difficult to implement in complex networks of variable pressure and strict water quality constraints. Recent efforts have introduced stochastic and metaheuristic approaches to solve large network optimization and combinatorial problems; however, their application for realetime control of water n et al., 2005]. supply and distribution systems is still limited [Bara Typical models include genetic algorithms, simulating annealing, and particle swarm optimization. These models are mostly based on iterative search algorithms that attempt problem solving with two or more, often conflicting, objective functions. Other techniques have been tested by water supply optimization practitioners pezeIb ~ ez, and include ant colony optimization [Ostfeld, 2008; Lo an 2008], honey bee mating optimization [Mohan, 2009], harmony search [Geem, 2009], tabu search [Cunha, 2004] and shuffled frog leaping optimization [Eusuff, 2003]. A critical review of these €rensen [2013], which identifies slight methods is presented by So differences in all evolution strategies that are based on the same underlying approach. Lately, hybrid approaches that are based on linear programming and genetic algorithms have been proposed and are built on the advantages of both deterministic and heuristic methods [Cisty, 2010]. In general, although considered ideal for optimizing energy and water quality management systems, the algorithms presented above have limited applications in real drinking water systems control. These techniques must oversimplify the network and the initial assumptions in order to overcome intrinsic challenges due to nonelinear network hydraulics and the subsequent advanced mathematical complexity and extensive fine tuning of parameters interferes with real‒time solutions.

4.3. Use of optimization techniques for water demand forecasting A key aspect of a successful EWQMS is the ability to accurately forecast water consumption and schedule production [Jentgen et al., 2003]. Water utilities have traditionally employed consumptionefollowing pumping operations in which wells and booster pumps are automatically controlled in response to reservoir levels and distribution pressures. In order to optimize pumping schedules, minimize energy or take advantage of energy supplier pricing schedules, a proactive rather than reactive system operation is needed. Therefore, the water demand forecaster is a critical functional component for an EWQMS that allows the leveraging of multiple energy management strategies to reduce costs by accurately forecasting consumption at 15e60 min intervals. For example, opportunities are identified to use storage to take advantage of TOU regimes, minimize simultaneous pumping operations and maximize operation of the most efficient pumps [Leiby and Burke, 2011; Jentgen et al., 2007]. While the water industry utilizes longeterm forecasting for infrastructure capital planning and shorteterm forecasting to set water rates, the extremely shorteterm daily or hourly forecasting utilized by the electric power and gas industries is a more suitable option for drinking water demand management and operations [Pacific Institute, 2013]. Current EWQMSs that are built on the basis of short-term demand forecasting do not have the capability to directly impact water demand and its management. Nevertheless, in conjunction with a robust hydraulic model, EWQMS may have the capability of identify some issues in the distribution system (e.g., leakage through pressure drops) that can be beneficial for water demand management and related strategies. Water utility extremely shorteterm consumption forecasting includes modeling with conventional regression methods and time series analysis, artificial intelligence techniques of expert systems and artificial neural networks (ANN) [Jain and Ormsbee, 2002]. Modeling techniques such as heuristic models, ANN, and regression models often show similar performances from houretoehour forecasts and their accuracies depend upon the precision and repeatability of the SCADA data, as well as the sophistication, calibration and level of maintenance of software tools [Jentgen et al., 2007]. Absolute relative errors ranging from 6.0% to 8.5%

116


Table 3 Typical shorteterm consumptive forecast (STCF) models used by water utilities [Jentgen et al., 2007]. Forecasting technique

Forecasting principle

Selected utilities using these techniques

Regression

Multievariable linear equations are developed to relate inputs (e.g., yesterday's consumption, current day rainfall, current day maximum temperature, current day solar radiation, etc.) to outputs (e.g., today's forecasted consumption). A forecast is made by searching from a historical database of consumption, weather data, etc. for the day that closely matches with the projected conditions of upcoming day. The base consumption is overlaid with longeterm trends for growth, seasonal consumption, regressive component that accounts for weather in order to make a water consumption forecast. The forecasting is generated from water system operator's heuristic knowledge of consumption behavior (e.g., change in water consumption relative to a threshold value) under different conditions (e.g., winter, summer, rainy periods). The model is trained by presenting it with a historical data set of observed inputs and the resulting observed consumption and an interactive process is used to minimize the error between the model prediction and observed consumption. A combination of two or more methodologies (regression, expert system, and ANN) into one larger application.

Method is used in conjunction with other methods presented below.

Similar Day

Time Series

Expert System (Rule Based)

Artificial Neural Network (ANN)

Hybrid

a b c d

Colorado Springs Utilitiesa

El Paso Water Utilities Dallas Water Utilities Lexington Kentucky American Water Company

Jacksonville Electric Authoritya San Diego Water Departmenta Las Vegas Valley Water Districta Toronto Waterb,c Washington Suburban Sanitary Commissionb,c Las Vegas Valley Water Districtb,c East Bay Municipal Utility Districtc,d Greater Vancouver Regional Districtc,d Seattle Public Utilitiesc,d

Installed STCF operational system. Regression, ANN, Heuristic. STCF prototypes. ANN, Heuristic.

for the hourly heuristic prototype, 5.8%e8.4% for the ANN prototype, and 5.7%e5.9% for the linear regression prototype were reported; the accuracy generally lowers when switching from daily to hourly forecasts [Jentgen et al., 2007]. Recently, newer forecasting strategies have relied upon the implementation of ANN models, expert systems, and “fuzzy logic” deployed through commercial systems (e.g., Nostradamus). Table 3 summarizes the key features of this modeling for shorteterm forecasting and the utilities in which the given model has been implemented.

5. Benefits of implementing an EWQMS Water utilities with complex networks containing multiple pressure zones, a mixed water resource portfolio and highly variable energy and demand tariffs or peak penalties, should benefit from implementation of an EWQMS. Implementation of an EWQMS can be very complex and entail extensive resources and capital investment. So far, only large utilities that are able to sustain the financial burdens of an automated platform and are equipped with

sophisticated hardware and software have opted for EWQMS control. The highest share of the initial capital investment is reported to be associated with the upgrade of the SCADA and of the network telemetry and remote control (e.g., flow meters and reservoirs control levels, etc.). For some of the utilities listed in Table 4 the SCADA was upgraded prior to EWQMS implementation and was not part of the initial investment. The estimated project costs for EWQMS design and implementation (i.e., software license, implementation, configuration and personnel training) was $0.75M at EBMUD, $1.87M at EMWD, $1.3M at EPWU and $3.75M at NWL (Badruzzaman et al., 2014). Based on the savings achieved, the payback periods for EWQMS implementation at these utilities ranged from 2 years at NWL and up to 5 years for the implementation at EMWD. At EBMUD the implementation cost was repaid based on actual savings achieved over a four year period against a sliding scale. In addition to the cost of the optimization software and of the adequate infrastructure, utilities need to sustain annual expenses associated with maintenance fees (typically 8%e10% of the project cost), labor, training for operators, calibration of meters and software (Badruzzaman et al., 2014).

Table 4 Benefits of EWQMS implementation at selected water utilities [Bunn, 2006; Derceto, 2013; EMA, 2013]. Utility

Peak demand (MGD)

% of network under EWQMS control

Electricity cost reduction

Energy savingsa

East Bay Municipal Utility District (EBMUD) Eastern Municipal Water District (EMWD) El Paso Water Utilities (EPWU) Las Vegas Valley Water District (LVVWD) Washington Suburban Sanitary Commission (WSSC) WaterOne Region Of Peel Northumbrian Water Limited (NWL) Gwinnett County Department of Water Resources (GCDWR)

130 275 160 430 240 100 290 350 70

33% 100% 30% 100% 100% 100% 100% 96% 100%

13% 15% 10%b 11% 11% 14% 10% 5% 8%

6% 8% 6%b 6% 8% 6% 6% N/A 6%

N/A: Data not available. a Percentage of kWh saved per million gallons of water delivered in respect to the kWh/MG used as a baseline value. b Predicted.


In order to quantify the actual benefits of an EWQMS installation, a baseline needs to be developed to allow a comparison of pree and posteinstallation conditions. The establishment of a baseline usually covers a one to two year period of “normal’ system operations and it is not straightforward in a dynamic system experiencing evolving changes (e.g., energy tariff changes, infrastructure changes, etc.) [Thorstensen, 2007]. EWQMS may produce tangible and intangible benefits including cost savings and demand charge avoidance, environmental benefits and other intangible values (e.g., improved uniformity of operations). Cost savings following implementation of an EWQMS vary annually and are utilityedependent. Electricity cost reductions are in the range of 8% (GCDWR) to 15% (EMWD), and 6%e9% energy savings were achieved, as reported in Table 4. In addition to economic and environmental benefits due to energy cost or use savings, utilities that opt for EWQMS implementation can take advantage of other operational benefits, such as improvement of water quality (e.g., managing water turnover in tanks), improved uniformity of operations, and improved emergency response and maintenance planning. EWQMS can be employed as a system simulator for decision making that assists engineers and planners in understanding the impact of changes in population growth, demand patterns, energy market profile changes, and other operational parameters and facilities. In combination with a robust hydraulic model EWQMS also assists water utilities in monitoring the water balance, identifying water losses and reducing water leakage in distribution system, to promptly facilitate routine implementations of water loss reduction activities (e.g., new infrastructure, maintenance). This is of high importance for utilities that have restricted water supplies and are required to find an optimum economic balance between available water resources and water demand.

6. Challenges of implementing an EWQMS EWQMS implementation encounters many challenges which are specific to the system infrastructure, the system boundaries, embedded energy and geographical configuration [Raucher et al., 2008]. Although all EWQMS installations rely on similar structural features, each one needs to be tailored to the specific system characteristics, objectives and limitations or constraints of each installation site. So far, only large utilities that are able to sustain the financial burdens of an automated platform and are equipped with sophisticated hardware and software have opted for EWQMS control. In general, the EWQMS implementation progresses in stages, first by optimizing the operation in control areas of the distribution system that can be isolated, and then by progressively expanding to the full network once the predicted savings and return on investments have been validated. As an example, Table 4 shows the level of EWQMS implementation in selected municipalities. Some of the implementation challenges of EWQMS reported in the literature are: Cultural shift. The implementation of an EWQMS may be a difficult task in water facilities with conservative operations and management practices [Barry, 2007]. The assimilation of energy efficiency concepts and their integration with water quality and GHG reduction is possible; however it usually requires major cultural and organizational changes [Atoulikian et al., 2011]. Operational constraints. Each utility operates within a set of specific constraints. Often these constraints are set in excess of what can be practically achieved. The number, type and

117

complexity of these constraints will impact the convergence of the EWQMS solutions. Operations with too many constraints may not produce a solution in a reasonable timeframe for real‒ time system operations. New generation SCADA. Real‒time control of water supply and distribution systems is not possible without adequate telemetry or SCADA. Early generation SCADA systems may be unsuitable for integration with sophisticated EWQMS software and may not guarantee adequate system coverage. Interface with the control system. Real‒time operation using an EWQMS requires interfacing to an existing control system. If the existing SCADA system and telemetry system do not include a remote control capability for pumps and valves in the field, the utility will require substantial investment in developing control logics (e.g., control hardware and telemetry, software) that communicate with the SCADA system. Interaction between SCADA and EWQMS. For real‒time operation of the EWQMS, accurate and timely input of values to the EWQMS from the SCADA is important. The utilities will not receive the expected benefits if the inputs to the model are either inaccurate or not delivered in time to be effectively used in the optimization and the instruments are not properly calibrated. Instrumentation and control. A water distribution network without at least a core set of measuring and monitoring instruments (e.g., flow meter, power meter) and remote control capability is unlikely to be optimized and operated on a real‒ time basis. As a bare, minimum pump station flows, pump run status indication and pressures are required. Metering infrastructure and sensor networks have a critical role in implementing and operating EWQMS. The installation of smart electric metering at pump stations and measurement of pump efficiency in real-time improves accuracy of the optimization solution and reliability of EWQMS. For example, the use of sensors (e.g., multi-parameter water quality sensors) is particularly beneficial to ensure that water quality goals are met and to help the software in selecting/blending water sources of given water chemistry. Other sensors, such as those providing flows and pressure transient data through the distribution system, may bring additional benefits, particularly as an early warning system for water quality episodes, leaks, and main breaks, and identify loss or decrease in disinfectant residual in a contamination event. Operator interaction. Operators can access the software functionalities for everyday decision making using a user-friendly interface. Through this interface, operators can actively interact with the software and override the solutions. The overrides take place to tag pumps in and out of service, modify pre-set inputs such as season demand profiles, set reservoir target levels, and change production limits. Typical reasons for manual interventions include failure to meet consumer demand, unpredicted variations in weather or water demand, emergency situations and unplanned maintenance.

Overrides are often justified by the operators based on safety, reliability and legal liability concerns that arise before they have established sufficient confidence in the software‘s solutions. The newly automated operations can suffer from too many manual overrides particularly if there is resistance from the operations team towards installation of full system control automation. In general, an operator works within soft boundaries, while the software has more arduous constraints to maintain. Also, because of the inherent uncertainty and natural risk adversity, an operator

118


may reduce trust in the program over time and develop an increasing tendency to override the system. Security. The SCADA network is a critical component of plant monitoring and processes control. A SCADA failure can alter system performance to the extent that it negatively impacts public health and safety and/or results in economic loss for a water utility. Thus, water utilities generally will not lock the existing SCADA infrastructure to the new EWQMS software components. In general, most facilities that implement an EWQMS need to have a robust control system in place, including warm or hotestandby SCADA servers as part of the system architecture that are ready to respond to any critical hardware or communication failure.

7. Knowledge gaps and future research needs The EWQMS is still in an early phase of adoption by the water industry and additional efforts are ongoing to attain even greater integration of water and energy management practices, enhance the existing methodology and surmount remaining challenges. Additional research as identified below is needed to further advance the current state of knowledge on EWQMS: Understanding the tradeoffs inherent to water utility operations. Many important decisions require trade-offs between conflicting objectives [Wu et al., 2011]. For many water authorities, multi-objective decisions on key operations and system planning need to be made considering uncertainties in future water supply and distribution management and increasing energy costs (Farmani et al., 2006). Despite this need for multi-objective analysis and optimization, specific tools are not widely available or used by the water industry for system planning and operations. It is important for water utilities to develop and apply modeling approaches to solve the existing conflicts between energy and asset management, water quality and overall reliability of water system operations. Framework to develop a business case. The cost of the software packages is just a fraction of total cost of an EWQMS. A majority of the payback period comes from the costs associated with the other resources that are needed to implement the system (e.g., upgraded SCADA, inehouse modeling capabilities, remote control equipment, interfacing between the different software packages, need for additional storage tanks or reservoirs, dedicated skilled staff, data collection needs for accurate system configuration, etc.). The business case for a utility that already operates with sophisticated SCADA and remote control capabilities, high storage to peak demand ratios, and advanced modeling capabilities may be more apparent than it is for a utility without any of these resources. Therefore, future research should develop a framework or decision matrix that can be readily used by a water utility to assess a business case that is commensurate with their capabilities. Energy cost saving versus energy consumption reduction. Typical EWQMS strategies implemented around the world focus on energy cost reduction with more limited emphasis on energy consumption reduction. Such an approach may limit a water utility's ability to seek opportunities to avoid energy wastage and to identify energy efficient alternatives. Thus, further research is needed to identify the costs and benefits of EWQMS when the optimization is conducted based on the least energy consumption operation rather than the least energy cost operation. The tradeeoff between energy cost saving and energy consumption should be developed.

Longeterm performances of EWQMS. The published literature only documents the benefits of EWQMS observed for a limited number of utilities operated for a short duration. Water utilities are oftentimes impacted by watershed events (e.g., drought/rain cycles which impact source water portfolios), national and/or regional regulations (e.g., water quality, air quality issues), budgetary constraints and electric utility tariff structure changes that force them to make major operational changes from year to year. However, although some utilities have been under EWQMS operation for many years, limited information is available on longeterm performance and operation with steady conditions overtime (e.g., infrastructure, water availability, energy tariff offers, energy source, etc.). GHG module. Water utilities are increasingly interested in GHG emission reductions. Three different EWQMS strategies have been undertaken to reduce carbon footprints: implementation of energy efficient programs; utilization of natural gas powered engines; and increased use of renewable power onesite. The existing EWQMS framework does not include a GHG module that would aid water utilities in understanding the impacts of operational changes into their daily GHG emissions. Real‒time GHG emissions. The carbon intensity of electricity is dependent on the combination of generation processes utilized which is timeedependent [Stokes et al., 2012]. The emission factors commonly employed in calculating the carbon footprint of water supply systems are regional and are typically expressed as a flat rate gCO2eeq/kWh regardless of time of use. Average emissions rates vary by season, but do not vary substantially between one and offepeak hours. In contrast, marginal emissions rates vary between one and offepeak hours, but do not vary substantially by season [Energy and Environmental Economics, 2009]. The variation of emission factors with respect to the electricity tariff structure may be of importance when optimizing pumping schedules. In fact the use of timeedependent emissions factors against more conventional static emission factors for pump operational management could result in further reducing operational GHG emissions [Stokes et al., 2012]. Joint water and electric utility programs. Bringing energy and water utilities together allow for greater saving opportunities and increased knowledge of the relationship between the water and energy resources at a water utility. The development of joint programs between the two sectors during the planning, operation and evaluation phase are recommended to make costeffective efficiency measures, increase savings of both energy and water resources and share the financial burdens between the two sectors (Young, 2013). Power utility incentive programs and benefit allocations can be developed to support Permanent Load Shifting (PLS) solutions, such as Aquadapt, at water utilities, particularly those with elevated storage capacities. Acknowledgments The authors gratefully acknowledge the Water Research Foundation and California Energy Commission (CEC)’s financial, technical, and administrative support (Project Funding Agreement 04271). We are particularly indebted to Linda Reekie in her role as the Foundation Project Officer and to the Project Advisory Committee of the Water Research Foundation. The support of David Weightman, contract manager at the California Energy Commission, is gratefully acknowledged. The comments and views detailed herein may not necessarily reflect the views of the Water Research Foundation and CEC, its officers, directors, employees, affiliates or agents. The authors would also like to thank the staff and Operations teams at East Bay Municipal Utility District, Eastern Municipal


Water District, El Paso Water Utilities, Las Vegas Valley Water District and Northumbrian Water Limited for their thoughtful contributions to this project. In conclusion, we would like to acknowledge the valuable contribution of Simon Bunn and Matthew Gordon, Derceto, Inc. and Dr. Paul Boulos of Innovyze, Inc. References Atoulikian, R., Ernest, J., Wellens, B., McGuckin, R., George, S., 2011. Implementing a sustainable energy efficiency improvement program at DC Water. In: Proceedings of the Water Environment Federation, Energy and Water 2011, pp. 1083e1093. Badruzzaman, M., Cherchi, C., Oppenheimer, J., Gordon, M., Bunn, S., Jacangelo, J.G., Implementation of energy and water quality management systems modified with a GHG module, In: Proceedings of the Annual Conference & Exposition '15, AWWA Conference, Boston, June 2014. Bagirov, A.M., Barton, A.F., MalaeJetmarova, H., Al Nuaimat, A., Ahmed, S.T., Sultanova, N., Yearwood, J., 2012. An algorithm for minimization of pumping costs in water distribution systems using a novel approach to pump scheduling. Math. Comput. Model. 57 (3e4), 873e886. Bakker, M., Verberk, J.Q.J.C., Palmen, L.J., Sperber, V., Bakker, G., 2011. Higher energy efficiency and better water quality by using model predictive flow control at water supply systems. In: Proceedings of the Water and Innovation: Water Technology Conference, International Water Week Amsterdam, November 2011. Bar an, B., Von Lücken, C., Sotelo, A., 2005. Multieobjective pump scheduling optimisation using evolutionary strategies. Adv. Eng. Softw. 36 (1), 39e47. Barnett, M., Lee, T., Jentgen, L., Conrad, S., Kidder, H., Woolschlager, J., Cantú Lozano, E., Kelly, S., Eaton, M., Hollifield, D., Groff, C., 2004. Realetime automation of water supply and distribution for the city of Jacksonville, Florida, USA. EICA 9 (3), 15e29. Barry, J.A., 2007. WATERGY: Energy and Water Efficiency in Municipal Water Supply and Wastewater Treatment, Costeeffective Savings of Water and Energy. The Alliance to Save Energy (Report). Beyer, D.V., Young, J.A., White, E.M., Bunn, S., 2005. Implementing an energy management system for distribution pumping. In: Proceedings of the American Water Works Association Annual Conference and Exhibition 2005, San Francisco, CA, USA, June 2005. Boulos, P.F., Lansey, K.E., Karney, B.W., 2006. Comprehensive Water Distribution Systems Analysis Handbook for Engineers and Planners, MWH Soft, 2006. Pennsylvania State University. Boulos, P.F., Bros, C.M., 2010. Assessing the carbon footprint of water supply and distribution systems. J. Am. Water Works Assoc. 102 (11). Brueck, T., Geisenhoff, J., 2010. Decision support system for sustainable energy management. In: Proceedings of Annual Conference and Exposition 2010 of the American Water Works Association, June 20e24, 2010, Chicago, IL. Bunn, S., 2006. Pump scheduling optimization in four US cities: case studies. In: Proceedings of the 8th Annual Water Distribution Systems Analysis Symposium, Cincinnati, Ohio. Bunn, S., 2011. Optimizing operations holistically for maximum savings. In: Proceedings of Annual Conference and Exposition 2010 of the American Water Works Association, June 20e24, 2010, Chicago, IL. Bunn, S., Ghaderi, K., Hillebrand, C., 2006. Utilizing energy and demand forecasting as a tool to develop energy procurement strategies. In: Proceedings of Annual Conference and Exposition 2006 of the American Water Works Association, DSS Section, September 2006, Phoenix, AZ, USA. Bunn, S.M., Reynolds, L., 2009. The energyeefficiency benefits of pumpescheduling optimization for potable water supplies. IBM J. Res. Dev. 53 (3), 5e1. Bunn, S., Hillebrand, C., 2008. Operating pumps to maximize efficiency and to minimize your carbon footprint or save money, save the planet. In: Proceedings of Annual Conference and Exposition 2008 of the American Water Works Association. Chaudhry, S., Shrier, C., 2010. Energy sustainability in the water sector: challenges and opportunities. In: Proceedings of Annual Conference and Exposition 2010 of the American Water Works Association, June 20e24, 2010, Chicago, IL. Cisty, M., 2010. Hybrid genetic algorithm and linear programming method for leastecost design of water distribution systems. Water Resour. Manag. 24 (1), 1e24. Cunha, M.D.C., Ribeiro, L., 2004. Tabu search algorithms for water network optimization. Eur. J. Operational Res. 157 (3), 746e758. Derceto, Inc. (2013), http://www.derceto.com/. EMA (2013), Available online at: http://www.ema-inc.com/success-stories/ operational-optimization. Energy and Environmental Economics, 2009. Meeting California's Long-term Greenhouse Gas Reduction Goals. Report, Available online at: http://www. ethree.com/documents/2050revised.pdf. Eusuff, M.M., Lansey, K.E., 2003. Optimization of water distribution network design using the shuffled frog leaping algorithm. J. Water Resour. Plan. Manag. 129 (3), 210e225. Farmani, R., Walters, G., Savic, D., 2006. Evolutionary multi-objective optimization of the design and operation of water distribution network: total cost vs. reliability vs. water quality. J. Hydroinformatics 8 (3), 165e179. Fowler, M.R., Main, M.H., 2010. Optimisation of energy in water supply using pump

119

scheduling. In: Proceedings of the IWA Water and Energy Conference, Amsterdam, 11 November, 2010. Geem, Z.W., 2009. Particleeswarm harmony search for water network design. Eng. Optim. 41 (4), 297e311. Goonetilleke, R., 2011. Hybrid Pumping: using a natural gas engine and electric motor, yields the lowest possible energy costs with the greatest reliability. In: Presented by the Southern California Gas Company and San Diego Gas & Electric, April 28, 2011. Griffiths-Sattenspiel, B., Wilson, W., 2009. The Carbon Footprint of Water. River Network, Portland. ndez, E., Pardo, M.A., Cabrera, E., Cobacho, R., 2010. Energy assessment of Herna water networks, a case study. In: Proceedings of the Water Distribution System Analysis 2010 Conference e WDSA2010, Tucson, AZ, USA, Sept. 12e15, 2010. Innovyze, Inc. (2013), Available online at: www.innovyze.com. Jain, A., Ormsbee, L.E., 2002. Shorteterm water demand forecast modeling techniques: conventional methods versus AI. J. Am. Water Works Assoc. 94 (7), 64e72. Jentgen, L.A., Conrad, S., Riddle, R., Sacken, E.V., Stone, K., Grayman, W., Ranade, S., 2003. Implementing a Prototype Energy and Water Quality Management System. AWWA Research Foundation (Report). Jentgen, L.A., Conrad, S., Kidder, H., Barnett, M., Lee, T., 2005. Optimizing Operations at JEA's Water System, American Water Works Research Foundation (Report). Jentgen, L.A., Kidder, H., Hill, R., Conrad, S., 2007. Energy management strategies use shorteterm water consumption forecasting to minimize cost of pumping operations. J. Am. Water Works Assoc. 99 (6), 86e94. Jowitt, P.W., Germanopoulos, G., 1992. Optimal pump scheduling in wateresupply networks. J. Water Resour. Plan. Manag. 118 (4), 406e422. Kanakoudis, V., 2014. Three alternative ways to allocate the cost of the CF produced in a water supply and distribution system. Desalination Water Treat. 1e11. Kanakoudis, V., Gonelas, K., 2014a. Accurate water demand spatial allocation for water networks modelling using a new approach. Urban Water J. 1e18 (aheadof-print). Kanakoudis, V., Gonelas, K., 2014b. Developing a methodology towards full water cost recovery in urban water pipe networks, based on the “user-pays” principle. Procedia Eng. 70, 907e916. Kanakoudis, V., Tsitsifli, S., Samaras, P., Zouboulis, A., 2013a. Assessing the performance of urban water networks across the EU Mediterranean area: the paradox of high NRW levels and absence of respective reduction measures. Water Sci. Technol. Water Supply 13 (4), 939e950. Kanakoudis, V., Tsitsifli, S., Samaras, P., Zouboulis, A., Banovec, P., 2013b. A new set of water losses-related performance indicators focused on areas facing water scarcity conditions. Desalination Water Treat. 51 (13e15), 2994e3010. Kanakoudis, V., Tsitsifli, S., Papadopoulou, A., 2012. Integrating the carbon and water footprints' costs in the water framework directive 2000/60/EC full water cost recovery concept: basic principles towards their reliable calculation and socially just allocation. Water 4 (1), 45e62. Kanakoudis, V., Tsitsifli, S., Cerk, M., Banovec, P., Samaras, P., Zouboulis, A.I., 2014. Basic principles of a DSS tool developed to prioritize NRW reduction measures in water pipe networks. Water Qual. Expo. Health 1e13. Kanakoudis, V.K., 2004. Vulnerability based management of water resources systems. J. Hydroinformatics 6 (2), 133e155. €rkka €inen, S., Ik€ Ka aheimo, J., 2009. Integration of demand side management with variable output DG. In: Proocedings of the 10th IAEE European Conference, September, 2009, Vienna, pp. 7e10. Klein, G., 2009. The WatereEnergyeGreenhouse gas connection. Proc. Water Environ. Fed. 2009 (7), 7907e7915. Kurek, W., Ostfeld, A., 2013. Multieobjective optimization of water quality, pumps operation, and storage sizing of water distribution systems. J. Environ. Manag. 115, 189e197. Leiby, V.M., Burke, M.E., 2011. Energy Efficiency Best Practices for North American Drinking Water Utilities. Water Research Foundation. Report, Available online at: www.nyserda.ny.gov/_and_/EEBP_drinking_water_utilities.pdf. pezeIba n ~ ez, M., Prasad, T.D., Paechter, B., 2008. Ant colony optimization for Lo optimal control of pumps in water distribution networks. J. Water Resour. Plan. Manag. 134 (4), 337e346. Mohan, S., Babu, K.J., 2009. Optimal water distribution network design with honeyebee mating optimization. J. Comput. Civ. Eng. 24 (1), 117e126. Morley, M.S., Bicik, J., VamvakeridoueLyroudia, L.S., Kapelan, Z., Savic, D.A., 2009. Neptune DSS: a decision support system for nearereal time operations management of water distribution systems. In: Boxall, Maksimovic (Eds.), Integrating Water Systems. Taylor and Francis Group, London, pp. 249e255. Murphy, L., McIver, D., Dandy, G.C., Hewitson, C., Frey, J.P., Jacobsen, L., Fang, M., 2007. GA optimization for Las Vegas Valley water distribution system operations and water quality. In: Proceedings of the World Environmental and Water Resources Congress 2007: Restoring Our Natural Habitat, 15e19 May 2007, Tampa, FL, USA. Nace, D., Demotier, S., Carlier, J., Daguinos, T., Kora, R., 2001. Using linear programming methods for optimizing the realetime pump scheduling, bridging the Gap: meeting the World's water and environmental resources Challenges. In: Proceedings of the 2001 EWRI, ASCE Conference. Nasirian, A., Maghrebi, M.F., Yazdani, S., 2013. Leakage detection in water distribution network based on a new heuristic genetic algorithm model. J. Water Resour. Prot. 5, 294e303. NYSERDA, New York State Energy Research & Development Authority, 2010. Water & Wastewater Energy Management Best Practices Handbook. NYSERDA.

120


Available online at: http://www.nyserda.ny.gov. Ostfeld, A., Tubaltzev, A., 2008. Ant colony optimization for leastecost design and operation of pumping water distribution systems. J. Water Resour. Plan. Manag. 134 (2), 107e118. Pacific Institute, 2013. Water Rates: Water Demand Forecasting. Available on line at: www.allianceforwaterefficiency.org. Patki, V.K., Shrihari, S., Manu, B., Deka, P.C., 2013. Fuzzy system modeling for forecasting water quality index in municipal distribution system. Urban Water J. 1e22 (ahead-of-print). Penny, M., 2011. Visible Intelligent Networks. Yorkshire Water. Available on line at: http://www.swan-forum.com/uploads/5/7/4/3/5743901/_penny_swan_2011. pdf. PLMA, Peak Load Management Alliance, 2002. Demand Response: Principles for Regulatory Guidance. Jupiter, Fla. Peak Load Management Alliance (Report). Rance, J., Coulbeck, B., Kosov, S., Bounds, P., Ulanicki, B., 2001. FINESSEea Comprehensive Software Package for Water Network Modelling and Optimization. Research Studies Press Ltd, Baldock, England. Raucher, R.S., Cromwell III, J.E., Cooney, K., Thompson, P., Sullivan, L., Carrico, B., MacPhee, M., Wilkinson, R., 2008. Risks and Benefits of Energy Management for Drinking Water Utilities. AWWA Research Foundation (Report). Reynolds, L.K., Bunn, S., 2010. Improving Energy Efficiency of Pumping Systems through Realetime Scheduling Systems, Integrating Water Systems e Boxall & Maksimovi'c Editions. Taylor & Francis Group, London. RMI, 2006. Demand Response: an Introduction e Overview of Lessons, Technologies, and Lessons Learned. Rocky Mountain Institute. Rossman, L.A., 1994. EPANET Users Guide, Drinking Water Research Division, Risk Reduction Engineering Laboratory, Office of Research and Development. US Environmental Protection Agency, Cincinnati, Ohio (Report). Rothausen, S.G., Conway, D., 2011. Greenhouse-gas emissions from energy use in the water sector. Nat. Clim. Change 1 (4), 210e219. Salomons, E., Goryashko, A., Shamir, U., Rao, Z., Alvisi, S., 2007. Optimizing the operation of the HaifaeA wateredistribution network. J. Hydroinformatics 9 (1), 51e64. Sanders, K.T., Webber, M.E., 2012. Evaluating the energy consumed for water use in the United States. Environ. Res. Lett. 7 (3), 034034. Senon, C., Badruzzaman, M., Contreras, A., Adidjaja, J., Allen, M., Jacangelo, J., 2013. Drinking Water Pump Station Design and Operation for Energy Efficiency. Water Research Foundation. Report, Available online at: www.waterrf.org/ Pages/Projects.aspx?PID¼4308. Skeens, B., Wood, W., Spivey, N., 2009. Water production and distribution realetime energy management. In: Proceedings of the AWWA DSS Conference (Reno, NV, USA, September).

Skworcow, P., AbdelMeguid, H., Ulanicki, B., Bounds, P., 2009. Optimal pump scheduling with pressure control aspects: case studies. In: Integrating Water Systems: Proceedings of the 10th International Conference on Computing and Control in the Water Industry. September, 2009. €rensen, K., 2013. Metaheuristicsdthe Metaphor Exposed, International TransSo actions in Operational Research (Early view online publication). Sovacool, B.K., Sovacool, K.E., 2009. Identifying future electricityewater tradeoffs in the United States. Energy Policy 37 (7), 2763e2773. Soyupak, S., Kilic, H., Karadirek, I., Muhammetoglu, H., 2011. On the usage of artificial neural networks in chlorine control applications for water distribution networks with high quality water. J. Water Supply: Res. TechnologydAQUA 60 (1), 51e60. Stokes, C., Maier, H.R., Simpson, A.R., 2012. Water distribution system greenhouse gas emissions reduction by considering the use of timeedependent emissions factors. In: Proceedings of the 14th Water Distribution Systems Analysis Conference. WDSA 2012, Adelaide, South Australia, September 2012. Thorstensen, S., 2007. Energy savings analysis generated by a real time energy management system for water distribution. In: Proceedings of the American Water Works Association Pennsylvania/Chesapeake Section Conference, Hershey PA, USA. Tucciarelli, T., Criminisi, A., Termini, D., 1999. Leak analysis in pipeline system by means of optimal value regulation. J. Hydraulic Eng. 125 (3), 277e285. United States Environmental Protection Agency (U.S. EPA), 2012. National Water Program 2012 Strategy: Response to Climate Change. December 2012. Wallis, M.J., Ambrose, M.R., Chan, C., 2008. Climate change: charting a water course in an uncertain future. J. Am. Water Works Assoc. 100 (6). Water Energy Innovation, 2013. The Role of Natural Gas in California's Watereenergy Nexus. Report, April 2013. Available online at: http://www. waterenergyinnovations.com/publication/the-role-of-natural-gas-incalifornias-water-energy-nexus/wppa_open/. Wilkinson, R.C., 2008. Integrating water and energy resource management in the context of climate change. In: The Proceedings of the International Conference on Water Scarcity, Global Changes, and Groundwater Management Responses. University of California, Irvine, United States. December 2008. Woodward, K., Fowler, M.R., 2011. Proof of concept of an oneline pump scheduling and control system. In: Proceedings of the CCWI Conference, Exeter and EWWM (London, UK September). Wu, W., Simpson, A.R., Maier, H.R., 2011. Sensitivity of optimal tradeoffs between cost and greenhouse gas emissions for water distribution systems to electricity tariff and generation. J. Water Resour. Plan. Manag. 138 (2), 182e186. Young, R., 2013. Saving Water and Energy Together: Helping Utilities Build Better Programs. American Council for an Energy-Efficient Economy (October).

Nanomaterials for energy and water management.

Water quality planning and management for Dapeng bay, China.

Medical laboratory quality systems - a management review.

Climate change, water quality, and water-related diseases in the Mekong Delta Basin: a systematic review.

Reducing nitrate loss in tile drainage water with cover crops and water-table management systems.

Integrating membrane filtration into bioelectrochemical systems as next generation energy-efficient wastewater treatment technologies for water reclamation: A review.

The Niagara River: A water quality management overview.

Water Quality Index for measuring drinking water quality in rural Bangladesh: a cross-sectional study.

Linking Spatial Variations in Water Quality with Water and Land Management using Multivariate Techniques.

Surface water quality management using an integrated discharge permit and the reclaimed water market.

Valuing flexibilities in the design of urban water management systems.

Cryptosporidium and Giardia in surface water: a case study from Michigan, USA to inform management of rural water systems.

Integrated hydrological and water quality model for river management: a case study on Lena River.

Are interactive effects of harmful algal blooms and copper pollution a concern for water quality management?

Risk factors and monitoring for water quality to determine best management practices for splash parks.

Intermittent Water Supply: Prevalence, Practice, and Microbial Water Quality.

Evaluation of water quality based on a machine learning algorithm and water quality index for the Ebinur Lake Watershed, China.

Variability in the chemistry of private drinking water supplies and the impact of domestic treatment systems on water quality.

Quality of drinking water.

Attenuation coefficients for water quality trading.

Water quality criteria for 2,4,6-trinitrotoluene.

Canopy management in rainfed vineyards (cv. Tempranillo) for optimising water use and enhancing wine quality.

Suspended sediment regimes in contrasting reference-condition freshwater ecosystems: implications for water quality guidelines and management.

Development of the Trophic Water Quality Index (TWQI) for subtropical temperate Brazilian lotic systems.