E363

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

Calculating the greenhouse gas emissions of water utilities ANTHony H. Johnston1 and Tanju Karanfil2 1Mustang

Engineering, Greenville, S.C. of Environmental Engineering and Earth Sciences, Clemson University, Anderson, S.C.

2Department

The objective of this study was to develop and test a userfriendly greenhouse gas (GHG) emissions accounting tool for water utilities. The tool was designed to assist utilities in calculating GHG emissions for the purposes of complying with future regulation and reporting requirements, providing information for city and state climate action plans, and fulfilling voluntary carbon emissions goals. For utilities lacking accurate energy use data, equations were developed to predict energy use

for three phases of water production: raw water collection, water treatment, and finished water distribution. The tool was tested by seven Georgia, North Carolina, and South Carolina utilities, which had an average carbon intensity of 1,240 kg carbon dioxide equivalents (CO2-eq)/mil gal when scope 1 and 2 emissions were evaluated. However, the averages ranged from 550 to 2,190 kg CO2-eq/mil gal, depending on the electrical grid used.

Keywords: carbon inventory, greenhouse gas emissions, sustainability

Recent increased strain on potable water sources, as well as concern about environmental issues such as climate change, has led to greater emphasis on sustainable practices (Vince et al, 2008). For water utilities, this has intensified scrutiny on energy use. Although traditionally viewed solely in financial terms, energy use is also the primary source of greenhouse gas (GHG) emissions from water utilities (Vince et al, 2008). Emerging concern about GHG emissions coincides with potential federal legislation and regulation by the US Environmental Protection Agency (USEPA; Hoffman, 2010). Water utilities may also have an interest in reducing GHG emissions and the effects of climate change beyond regulatory compliance. Climate change is decreasing both the quantity and quality of the water sources available to some utilities while increasing the demand for potable water that occurs with rising temperatures (Wallis et al, 2008; Goldstein & Smith, 2002). The GHG emissions from water utilities may also be a considerable contributor to the carbon inventory of individual cities. Water production accounted for 31% of total emissions from governmental operations in a study of 2006 data from Columbia, S.C. (Brennan, 2011). In order to determine their GHG emissions, water utilities need readily available guides and tools. Information to educate water utilities about their GHG emissions is often scattered, and calculation tools designed specifically for water utilities are not publicly available in the United States. This article presents a simple and straightforward way for water utilities to calculate their GHG emissions with a user-friendly GHG emissions accounting tool.

BACKGROUND GHG emission concepts. When the emissions of water utilities are evaluated, the specific GHGs of interest are carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O; Huxley et al, 2009). To analyze GHG emissions as a single value, a single unit of measure called carbon dioxide equivalents (CO2-eq) is used. To convert the amount of an individual GHG emission to CO2-eq, the global warming potential (GWP) of that particular GHG is used (Huxley et al, 2009). GWP compares the warming effect of GHGs to that of CO2. Thus, 1 kg of CH4 with a GWP of 21 has the same warming effect as 21 kg of CO2. The GWP values of the three previously mentioned GHGs are shown in Table 1. When GHG emissions are accounted for, three categories have been established to identify the source of the emissions (WRI, 2004). These categories are known as scope 1, scope 2, and scope 3. Scope 1 (direct emission sources) represents emissions from onsite and mobile combustion. Scope 2 (indirect emission sources) represents emissions caused by production of the electricity, steam, and hot or chilled water used by the water utility. Scope 3 represents indirect emission sources over which the utility has a measurable amount of control. Common scope 3 emission sources are employee commuting and business travel. Examples more closely related to water utilities are the emissions associated with the production and transport of chemicals used at the treatment plant. Life cycle assessment studies. Life cycle assessment (LCA) studies were used to identify the major sources of environmental effects, specifically the GHG emissions of water utilities. An LCA

2013 © American Water Works Association

E364

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

TABLE 1

oil, natural gas, wind, nuclear energy, and so on). Each of these energy sources has a different GHG emissions factor—i.e., the GHG emissions per kilowatt hour of electricity produced (Figure 1). Thus, the cleaner the energy supplied, the lower the water utility’s GHG emissions.

GWP values for various GHGs

Common Name Carbon dioxide

Chemical Formula

GWP100*

CO2

1

Methane

CH4

25

Nitrous oxide

N2O

298

CALCULATing A WaTER UTILITY’S GHG EMISSIONS

GHG—greenhouse gas, GWP—global warming potential, GWP100—the number of years over which the global warming potential was calculated *Huxley et al, 2009

involves what is commonly referred to as a cradle-to-grave approach, in which the entire life cycle of a process, including all inputs and outputs, is taken into account (Vince et al, 2008; Raluy et al, 2005). The LCA studies resulted in three conclusions: (1)  The operational phase of a water utility’s life cycle is overwhelmingly the leading contributor (often > 90%) to environmental impact; the construction and decommissioning phases can be ignored even in relatively extreme conditions associated with large infrastructure requirements (Raluy et al, 2005). (2)  Within the operational phase, energy use and, more specifically, the production of electricity, is the greatest source of environmental impact (Vince et al, 2008). (3)  After energy use, the chemicals used for water treatment represent the second largest source of environmental impact (Vince et al, 2008). Importance of the electrical grid. Water utilities obtain electricity from their local electrical grid. The power on that grid is produced at one or several generating stations, each of which may use a different source of energy to generate electricity (e.g., coal,

FIGURE 1

The process for calculating the GHG emissions of a water utility is divided into several phases: electricity source, raw water collection, treatment process, finished water distribution, and other energy uses, such as those by buildings and fleet vehicles. Because annual assessments are most common, annual values are used for calculations in the GHG emissions accounting tool developed in this study. A copy of the tool can be downloaded from the website of Clemson University’s Department of Environmental Engineering and Earth Sciences (www.clemson.edu/ ces/eees/outreach/index.html). Detailed information about the tool’s development can be found in the accompanying thesis (Johnston, 2012), available on the same website. Electrical grid. The first step is to determine the GHG emission factors to use for the utility’s purchased electricity. Three common options for selecting these factors are listed here in decreasing order of accuracy (option 1 being the most accurate and option 3 being the least accurate). Option 1 is to obtain the applicable GHG emission factors (pounds or kilograms of CO2-eq/megawatt hour) directly from the utility’s electricity provider. Option 2 is to use the emission factors provided by USEPA for the subregion in which the utility is located. The USEPA has divided the United States into 26 subregions, which represent more localized electrical grids than the grid for the nation as a whole. The subregions and their corresponding emission fac-

Average life-cycle GHG emissions resulting from various methods of producing electricity*

Average Emissions—g CO2-eq/kW·h

1,200 1,000 800 600 400 200 0 Coal

Natural Gas

Oil

Nuclear

Hydroelectric

Electricity Production Method CO2-eq—carbon dioxide equivalent, GHG—greenhouse gas *Johnston, 2012

2013 © American Water Works Association

Biomass

Wind

Solar

E365

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

tors can be found in USEPA’s eGRID tool (USEPA, 2010). Option 3 is to use the national average emission factors. Raw water collection. The major source of GHG emissions from raw water collection is the energy used to pump water. Pumping involves scope 1 emissions from fuel use and scope 2 emissions from electricity use. Once determined, annual electricity use for raw water collection is multiplied by the emission factors from the electrical grid obtained earlier. Similarly, the amount of fuel used annually is multiplied by the emission factor for that specific fuel. Emission factors for various fuels (e.g., natural gas, propane, diesel, and so on) are also available (Johnston, 2012). Water treatment. Water treatment processes involve scope 1 emissions from fuel use, ozone generation with air, onsite carbon reactivation, natural processes within standing bodies of water, and sludge disposal; scope 2 emissions from the electricity used during treatment; and scope 3 emissions from the production and transport of treatment chemicals used by the utility. As with the raw water collection phase, the first data to be collected are the annual amounts of electricity and fuel used during water treatment. Again, these amounts are multiplied by the corresponding emission factors to determine annual GHG emissions. The next type of data to be collected for the treatment process phase is scope 1, or direct emission sources, that are specific to the process of producing potable water. First, if ozone is used for oxidation or disinfection, the annual volume of ozone generated should be determined. This would apply only to utilities that use air, not pure oxygen, to supply the ozone generation systems. The reason for this is that the nitrogen in the air reacts in the ozone generation process to produce N2O (Huxley et al, 2009). Next, data are collected on the amount of granular activated carbon annually regenerated onsite—i.e., within the organizational boundary of the water utility. This accounts for the carbon that is oxidized to CO2 and emitted to the atmosphere during the regeneration process. The tool described in this article assumes

TABLE 2

Direct GHG emission sources and emission factors associated with potable water production

that 7.5% of the regenerated carbon is released as CO2 (Huxley et al, 2009). This calculation does not include the fuel used to operate the regeneration process, which should be taken into account as part of annual fuel use during the treatment phase. The third category of data to be collected for the treatment phase involves reservoir GHG emissions caused by natural chemical and biological processes. The water utility will determine the surface area of its reservoirs and the type of climate in which the reservoirs are located (boreal, temperate, subtropical, or tropical). The last direct emissions source specific to potable water production involves GHG emissions from sludge disposal at a landfill. The annual amount (in tons) of total organic carbon that is removed and sent to a landfill must be calculated. The emission factors that are used to determine GHG emissions from these specific treatment processes are shown in Table 2. The last type of data to be collected for the treatment phase concerns the use of chemicals, which are included in scope 3 emissions. The annual amount (in pounds) of each chemical used in the treatment process is multiplied by the emission factor for the corresponding chemical (Table 3). Finished water distribution. The distribution of finished water involves scope 1 emissions from fuel use and scope 2 emissions from electricity use. The finished water distribution phase is like the raw water collection phase in that only energy use is of concern. Once determined, the data on annual electricity and fuel use are multiplied by the matching emission factors to determine annual GHG emissions. Buildings, fleet vehicles, and other energy uses. The last aspect of the water utility to be evaluated involves emission sources such as administrative buildings and utility-owned vehicles. This phase involves scope 1 emissions from fuel use by utility-owned vehicles and building heating and air-conditioning systems and scope 2 emissions from the electricity used by other aspects of utility operations not already considered. Again, data on annual electric-

TABLE 3

Chemical

Emission Factor* Source

Carbon Dioxide

Methane

Ozone generation GAC regeneration

Nitrous Oxide 0.11 g

N2O/m3

[(44/12)* 7.5%]/ton

Boreal

Subtropical Tropical Sludge disposal

1,460

mg/m2/d

525 mg/m2/d

57.2

mg/m2/d

6.7 mg/m2/d

Alum

276

Ferric chloride

77

Ferrous chloride

77

1,065

0.0 mg/m2/d

Lime

1,264

Polymers

2,082

0.2

6.7 mg/m2/d

0.0 mg/m2/d

5,470 mg/m2/d

136.1 mg/m2/d

218.8 mg/m2/d

762 kg/ton

39 kg/ton

780

Sodium hypochlorite

mg/m2/d

525 mg/m2/d

GAC—granular activated carbon, GHG—greenhouse gas *Huxley et al, 2009

GHG Emissions* kg CO2-eq/metric ton

Chlorine

Reservoir emissions

Temperate

GHG emission factors for chemical production

Carbon dioxide Oxygen

346 226

Sodium hydroxide

1,376

Ammonia

2,400

CO2-eq—carbon dioxide equivalents, GHG—greenhouse gas *Tripathi, 2007

2013 © American Water Works Association

E366

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

ity and fuel use are multiplied by the corresponding emission factors to determine annual GHG emissions. After the emissions from fuel and electricity use have been calculated, GHG emissions from mobile combustion sources must be determined. For scope 1 purposes, this includes any vehicle owned directly by the water utility—but not employee-owned vehicles. To determine the amount of CO2 emissions, only the volume of fuel used each year is required. Therefore, the annual amount of fuel used is multiplied by the specific CO2 emission factor for that fuel. The emissions of CH4 and N2O depend on more than simply the volume of fuel used. The USEPA emission factors rely on mileage, fuel type, vehicle type, and vehicle model year. The vehicle type and model year are required because CH4 and N2O emissions differ depending on the type of catalytic control in the vehicle. These control systems vary with vehicle type and model year, and newer cars often produce lower emissions as a result of improved catalytic control. The annual mileage for each vehicle will have to be determined, and multiplying the mileage data by the emission factors for the corresponding vehicle type, fuel type, and model year will determine that vehicle’s annual CH4 and N2O emissions. Emission totals. Once all the necessary GHG emissions data have been calculated, the emissions can be combined to provide an improved analysis. For example, emissions from the four phases can be analyzed along with the total for the entire water utility. The amount of each individual GHG (CO2, CH4, and N2O) can be evaluated along with combining them by using the GWP factors from Table 1 to determine the total CO2-eq. Most GHG protocols in the United States calculate, track, and report the sum of scope 1 and 2 GHG emissions. Most European GHG protocols calculate, track, and report the sum of scope 1, 2, and 3

FIGURE 2

GHG emissions. Many believe that in the near future, US GHG protocols will also include scope 3 emissions. Development of equations to predict energy use. Because energy use has been shown to be the single largest contributor to GHG emissions related to water production, water utilities that do not have detailed energy use data need a way to estimate their energy use. Although energy use studies have been conducted on both a state and national scale, the end results have simply been average values in kilowatt hours per thousand gallons (Figure 2). Although equations for predicting energy use have been developed in the past (Carlson & Walburger, 2007), the authors of this article believed improvements could be made. The energy use prediction equations described here can be useful for utilities that cannot separate their electricity use into the phases discussed but still want a closer analysis of their GHG emission sources. These equations can also be applied to estimate the energy use, and therefore the GHG emissions, associated with future projects. The goal in developing the equations was to create the ability to predict the energy use related to three phases of water production: raw water collection, treatment, and finished water distribution. To construct these equations, data on energy use and water system characteristics were initially obtained from a survey conducted by others (Carlson & Walburger, 2007). To augment this data set, a new survey—the Clemson survey—was designed and conducted for this study. The Clemson survey resulted in 37 new data sets to be added to existing data from the Carlson & Walburger data set. To sort through all the survey responses, two initial filters were used. Water utilities that did not provide information on flow rate and electricity use were deleted from further analysis. This action left 192 remaining

Results of previous energy use investigations*

12

Energy Use—kW·h/1,000 gal

11 10 9 8 7 6 5 4 3 2 1 0

Southern California

Northern California

Iowa

Wisconsin Location

*EPRI, 2009 †Carlson & Walburger, 2007

2013 © American Water Works Association

National Average†

Massachusetts

E367

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

FIGURE 3

A

Survey responses according to geographic distribution (A) and distribution based on utility size (B)

7

1 1

2 4

2

1 3 3

6

11

1

3

4 2

2

3

4 3

2

1

10

8

8 3

33

4

4

1 34

5

1

7 2

2 4

B

Number of Survey Responses

70 60 50 40 30 20 10 0

100

Utility Size—mgd

FIGURE 4

Scope 1 and 2 carbon intensities of the water utilities that tested the GHG emissions accounting tool

Utility A Utility E

Utility B Utility F

Utility C Utility G

IMPLEMENTing THE GHG EMISSIONS ACCOUNTING TOOL

Utility D Average

Scope 1 and 2 Carbon Intensity— kg CO2-eq/mil gal

1,800 1,600 1,400 1,200 1,000 800 600 400 200 0

0

10

20

30

40

50

60

70

Total Average Flow Rate—mgd CO2-eq—carbon dioxide equivalent, GHG—greenhouse gas

80

responses for further analysis. The geographic distribution of the survey responses is shown in Figure 3, part A, and the utility size distribution is shown in Figure 3, part B. To see the data collected and used to develop the equations for predicting energy use, refer to Johnston (2012). To develop the energy use–prediction equations, the statistics program Statistical Analysis Systems (SAS) was used because it allows the simple production of multiple linear regression models. SAS also has the ability to identify significant independent variables using the lasso selection method recommended by Flom and Cassell (2007) as an alternative to the more common forward-selection method. The first step was to identify possible independent variables that affect energy use and that were reported in both surveys for each of the three phases of water production. SAS was then used to run the lasso selection method to identify the significant independent variables, and this was followed by compilation of a regression model using those independent variables. The output regression model was then analyzed by means of various statistical tests. After any outlier or erroneous data points had been identified and deleted, the process of lasso selection, regression model formation, and model analysis was repeated until an acceptable regression model was obtained. One additional step was critical to developing the equations. Because the survey data had a large range and variance, a transformation of the data was required. Both a log, specifically log10, and a square-root transformation were investigated. The transformations were performed on both the dependent and independent variables, so the independent variable changed from energy use to the log10 of energy use. The final energy-use prediction equations are shown in Table 4. To include a method of predicting electricity demand for water treatment, data from the literature were selected (Table 5). Utility personnel can select the processes used at their utility to estimate the electricity demand per 1,000 gal. The utility’s total average flow rate (in million gallons per day) can then be used to estimate total electricity demand.

90

The GHG emissions accounting tool was implemented and tested at seven water utilities in Georgia, North Carolina, and South Carolina. These utilities provided input data for the program as well as feedback on the tool itself. To determine GHG emissions from electricity use, emission factors from the corresponding USEPA subregion were used for all seven utilities. Each utility’s treatment process, total average flow rate, and any unique characteristics that might affect GHG emissions are shown in Table 6. The best way to compare the water utilities with one another is to use the combined scope 1 and 2 carbon intensities. The reason for this is that not every chemical used at the various utilities was available in the GHG emissions accounting tool, leaving the possibility that one utility might have a larger scope 3 emissions value than could have been calculated. Figure 4 provides a visual comparison of the carbon intensities of the seven utilities. The average scope 1 and 2 carbon intensity of the utilities that tested the GHG emissions accounting tool was

2013 © American Water Works Association

E368

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

TABLE 4

Final equations for predicting energy use Phase of Treatment

Energy Use Prediction Equation

Raw water collection (without purchased water flow)

Log10 (electricity [kW·h/y]) = 3.04430 + 0.42367 × Log10 (total average flow [1,000 gpd])    + 0.57216 × Log10 (raw water collection pumping hp +1)

Raw water collection (with purchased water flow)

Log10 (electricity [kW·h/y]) = 2.91331 + 0.80696 × Log10 (total average flow [1,000 gpd])    + 0.51377 × Log10 (raw water collection pumping hp + 1)    – 0.35124 × Log10 (average purchased water flow [1,000 gpd])

Treatment process

No useful correlation was determined.

Finished water distribution

Log10 (electricity [kW·h/y]) = 3.6538 + 0.4259 × Log10 (total average flow [mgd])    + 0.6590 × Log10 (finished water distribution pumping hp + 1)

1,240 kg CO2-eq/mil gal. Although seven data points allow only limited analysis, Figure 4 shows that a few observations were possible. First, this data set seems to have no economyof-scale effect because the smallest and largest utilities in terms of flow actually have fairly similar scope 1 and 2 carbon intensities. This finding contrasts with most literature sources that describe an economy-of-scale effect in which utilities that produce larger quantities of water have lower energy use and therefore lower carbon intensities (Malcolm Pirnie, 2008). However, the seven data points in this study constitute too small a sample to confirm or deny findings from the literature on the topic of economies of scale. A hypothesis can be posited as to why the three lowest scope 1 and 2 carbon intensities were from utilities D, E, and G. Utility D explained that it strives to be as progressive and environmentally friendly as possible, and these goals would lead to decreased energy use and thus a lower carbon intensity value. The fact that utility E did not include any data on fuel use or emissions from fleet vehicles may have caused its lower carbon intensity. Because utility G operates completely as a wholesaler, it lacks the GHG emissions from a large distribution system and vehicle fleet that most of the other utilities have. The only literature data found for comparison with the test results came from LCA studies. Because the LCA studies included more than what is measured by scope 1 and 2 carbon intensities, the sum of all three scopes was the more appropriate value to compare. Figure 5 shows a comparison of data from the seven utilities with data from the literature. The literature data are shown as horizontal lines instead of data points because the LCA studies did not provide flow-rate data but rather GHG emissions based on a functional unit. The literature data represent two extremes of scope 1, 2, and 3 carbon-intensity values in that Friedrich (2002) took into account only the treatment plant, and Stokes and Horvath (2006) analyzed importing water over long distances. Given these limitations, it was encouraging that the carbon intensity values of all of the utilities evaluated fell between the limits of the literature values. Another comparison that is important for water utilities to make is with the USEPA’s GHG emissions-reporting rule, which requires reporting from organizations that have annual scope 1 emissions

TABLE 5

Treatment steps available for selection in the energyprediction tool and their electricity demand* Treatment Step

Energy Use kW∙h/1,000 gal

Static mixer

0.00475

Rapid mixer

0.0345

Flocculator (vertical)

0.0012

Flocculator (horizontal)

0.0027

Conventional sedimentation

0.0015055

Plate settlers

0.00168

Superpulsator

0.0385

Ballasted flocculator

0.0635

Solids contact clarifier

0.11

DAF (high-rate)

0.11

Gravity filtration

0.0315

Pressure filtration

0.08

MF/UF (encased)

0.25

MF/UF (submerged)

0.265

UV-LPHO

0.0105

UV-MP

0.0555

UV/AOP

0.385

Ozonation

0.060

Hypochlorite (onsite generation)

0.066

Decarbonators

0.155

Brackish water RO

3.1

Wastewater RO

2.55

Seawater RO

8.25

Thermal desalination

21

DAF—dissolved-air flotation, MF/UF—microfiltration/ultrafiltration, RO—reverse osmosis, UV/AOP—ultraviolet light/advanced oxidation process, UV-LPHO—ultraviolet light with lowpressure high-output lamp, UV-MP—ultraviolet light with medium-pressure lamps *Veerapaneni et al, 2011

> 25,000 metric tons of CO2-eq (USEPA, 2012). Water utilities are exempt from this rule, and all seven of the utilities tested had scope 1 emissions < 25,000 metric tons of CO2-eq/year. However, if USEPA expanded the reporting rule to include scope 2 emissions,

2013 © American Water Works Association

E369

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

TABLE 6

Descriptive information for utilities that tested the GHG emissions accounting tool Total Average Flow mgd

Utility

Unique Characteristic Affecting GHG Emissions

Treatment Process

Utility A

 4.2

Conventional with onsite hypochlorite generation

None

Utility B

83.4

Conventional with onsite hypochlorite generation

None

Utility C

74.7

Direct filtration with ozonation

None

Utility D

55.4

Conventional

Self-described as environmentally friendly

Utility E

60.3

Conventional

No data included for energy use by fleet vehicles

Utility F

34.0

Conventional

None

Utility G

17.5

Conventional

Operates as wholesaler

GHG—greenhouse gas

two of the utilities tested would be affected because they have annual GHG emissions greater than the reporting threshold. Utilities B and C exceeded the limit with emission values of 49,700 and 40,200 metric tons of CO2-eq/year, respectively. If an average carbon intensity of 1,240 kg CO2-eq/mil gal is assumed, water utilities with a flow rate > 55.2 mgd would have annual GHG emissions greater than the limit specified in USEPA’s current reporting rule. The electrical grid that is being used can have a significant effect on both the average carbon intensity and the average flow rate that would exceed the threshold in USEPA’s reporting rule. To analyze this effect, the GHG emissions of the seven utilities tested were also compared with the emission factors from the three USEPA subregions with the highest and lowest GHG emissions as well as the average national emission factors. Results of this analysis are shown below in Table 7.

TABLE 7

Average scope 1 and 2 carbon-intensity values for the utilities tested and the flow rates that would exceed USEPA’s reporting rule limit* Scope 1 and 2 Carbonintensity Values kg CO2-eq/mil gal

Flow Rate That Would Exceed USEPA’s Reporting Rule Limit mgd

WECC Rockies

2,190

 31.3

SPP North

1,805

 37.9

SERC Midwest

1,786

 38.3

National Grid

1,311

 52.2

WECC California

 698

 98.1

NPCC Upstate New York

 698

 98.1

ASCC Miscellaneous

 554

123.5

USEPA Subregion

ASCC—Alaska Systems Coordinating Council, CO2-eq—carbon dioxide equivalents, NPCC—Northeast Power Coordinating Council, SERC—Smithsonian Environmental Research Center, SPP—Southwest Power Pool, USEPA—US Environmental Protection Agency, WECC—Western Electricity Coordinating Council *Emission factors from the three USEPA subregions with the highest and lowest greenhouse gas emissions, as well as the average national emission factors, were used in these calculations.

Data from each utility can be used to illustrate the source of its GHG emissions—electricity, onsite combustion, or fleet fuels. However, only utility G had the data required to illustrate its GHG emissions by phase. The relative amounts of GHG emissions from each phase are illustrated in Figure 6, which shows GHG emissions from all three scopes. The major contributor to the scope 1 and 2 carbon intensity of utility G is pumping because raw water collection and finished water distribution account for 75% of its emissions. Utility G operates solely as a wholesaler, and it is expected that finished water distribution would account for more emissions at utilities that operate larger distribution networks. When the relative amounts of GHG emissions among scopes 1, 2, and 3 were considered, all the utilities tested could be evaluated. In a comparison of GHG emissions from all three scopes, scope 2 emissions dominated, making up at least 80% of the carbon footprint.

TABLE 8

Test results for the energy-use-prediction equations

Utility Phase

Utility

Actual Electricity Use kW∙h/year

Raw water collection

Utility A

 1,330,000

1,172,000

–12

Raw water collection

Utility B

24,098,000

15,387,000

–36

Raw water collection

Utility F

 6,798,000

 7,912,000

16

Raw water collection

Utility G

 4,538,000

 6,684,000

47

Treatment

Utility G

 2,306,000

  668,000

–71

Treatment and distribution

Utility A

 3,112,000

  656,000

–79

Treatment and distribution

Utility B

47,579,000

15,818,000

–67

Treatment and distribution

Utility F

25,490,000

 8,827,000

–65

Distribution

Utility G

3,953,000

3,209,000

–19

Overall utility operations

Utility D

30,360,000

22,448,000

–26

Overall utility operations

Utility E

41,531,000

29,333,000

–29

2013 © American Water Works Association

Predicted Electricity Use kW∙h/year

Difference %

E370

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

FIGURE 5

Comparison of the scope 1, 2, and 3 carbon intensities of utilities A–G with data* from the literature

Utility data Friedrich, 2002 Stokes and Horvath, 2006

Scope 1, 2, and 3 Carbon Intensity— kg CO2-eq/mil gal

2,500

2,000

1,500

1,000

500

0

0

10

20

30

40

50

60

70

80

90

Total Average Flow Rate—mgd CO2-eq—carbon dioxide equivalent *Adapted from Friedrich, 2002, and from Stokes & Horvath, 2006

FIGURE 6

Relative amounts of GHG emissions from each phase of utility G operations

Raw water collection Treatment Distribution Buildings, fleet, other

GHG–greenhouse gas

As data for evaluating the GHG emissions accounting tool were collected, data were also gathered to test the energy-use-prediction equations developed in this study. The results, shown in Table 8, do not include utility C because of the lack of required information. Testing was done for separate phases of each water utility whenever possible. If a utility was not able to separate electricity use for treatment and electricity use for distribution, for example, that combination was compared with the combined prediction results for those two phases. Although the sample size is small, it is noteworthy that the predictions for the treatment phase or treatment and distribution phases were low by 60–80%. The prediction function for the treatment phase appeared to be the weakest because of its inability to fit a regression model and because of the program’s reliance on a single source of literature data for predictions. The results somewhat confirmed this conclusion. One limitation of the literature data used in the program is their lack of data on emissions associated with handling solids or sludge. This could explain some of the difference in the predictions because solids-handling equipment can include large energy users such as thickeners (e.g., centrifuges) and additional pumping. A larger data set would be required to more accurately evaluate the performance of the energy-use-prediction equations. Even with a larger data set and improved models, however, there is a limit to how accurately energy use can be predicted. Therefore, using detailed, measured energy data is the best approach for obtaining more reliable results from any GHG accounting tool.

SUMMARY In order to identify the major sources of water utility GHG emissions, findings from previous LCA studies were used: •  The operational phase of a water utility’s life cycle is the overwhelming contributor to environmental impact, and construction and decommissioning phases can be ignored. •  Fuel use and purchased electricity are the greatest sources of GHG emissions within the operational phase. •  Chemicals used during treatment (scope 3 emissions) represent the second largest source of GHG emissions. A process for calculating a water utility’s GHG emissions was presented. Data to be used in the calculations were included or referenced. Detailed, measured energy data are essential for obtaining reliable results from any GHG accounting tool. However, to help utilities evaluate their GHG emissions when energy use information is lacking, equations for predicting energy use were developed for the three phases of water utility operations—raw water collection, treatment, and finished water distribution. A GHG emissions accounting tool designed for water utilities in the United States was also created. The tool, which uses electricity, fuel, chemical, and operational data to calculate a utility’s annual GHG emissions, was tested by seven utilities in Georgia, North Carolina, and South Carolina. The average scope 1 and 2 carbon intensity for the utilities tested was 1,240 kg CO 2-eq/mil gal; however, the averages varied from 550 to 2,190 kg CO2-eq/mil gal, depending on the electrical grid used. The GHG emissions accounting tool is available for download, along with an accompanying thesis, on the website of Clemson University’s Department of

2013 © American Water Works Association

E371

Johnston & Karanfil | http://dx.doi.org/10.5942/jawwa.2013.105.0073 Journal - American Water Works Association Peer-Reviewed

Environmental Engineering and Earth Sciences (www.clemson. edu/ces/eees/outreach/index.html).

REFERENCES

RECOMMENDATIONS

Carlson, S.W. & Walburger, A., 2007. Energy Index Development for Benchmarking Water and Wastewater Utilities. Awwa Research Foundation, California Energy Commission, & New York State Energy Research and Development Authority. AwwaRF, Denver.

Future work with the GHG emissions accounting tool should be directed at increasing the tool’s accuracy and usability. Emissions calculated with the use of emission factors for the relevant USEPA subregion and emission factors obtained directly from electricity providers should be compared. This comparison would give water utilities an idea of the benefits of obtaining the additional information. Further options for increasing the accuracy of the accounting tool include keeping the USEPA subregion emission factors updated, adding emission factors for treatment chemicals not already included, and improving energy-use predictions for the treatment phase. The usability of the accounting tool can be enhanced by a transition into a web-based program. This would increase accessibility and allow a data collection function to improve future prediction models. Another potential improvement would be to combine the tool with a similar one designed for wastewater utilities. This would provide a comprehensive tool for combined utilities and municipalities.

ACKNOWLEDGMENT

Brennan, A.F., 2011. Baseline Emissions Inventory Report. City of Columbia Climate Protection Action Campaign, Columbia, S.C.

EPRI (Electric Power Research Institute), 2009. Program on Technology Innovation: Electric Efficiency Through Water Supply Technologies—A Roadmap. EPRI, Palo Alto, Calif. Flom, P.L. & Cassell, D.L., 2007. Stopping Stepwise: Why Stepwise and Similar Selection Methods Are Bad, and What You Should Use. Proc. NorthEast SAS Users Group 2007 Conference, Burlington, Vt. Friedrich, E., 2002. Life-Cycle Assessment as an Environmental Management Tool in the Production of Potable Water. Water Science & Technology, 46:9:29. Goldstein, R. & Smith, W., 2002. Water and Sustainability (Vol. 4): U.S. Electricity Consumption for Water Supply and Treatment—The Next Half Century. EPRI, Palo Alto, Calif. Hoffman, J., 2010. GHGs: Legislative Update and What It Means to You. Proc. AWWA 2010 Ann. Conf., Chicago. Huxley, D.E.; Bellamy, W.D.; Sathyanarayan, P.; & Ridens, M., 2009. Greenhouse Gas Emission Inventory and Management Strategy Guidelines for Water Utilities. Water Research Foundation & AWWA, Denver. Johnston, A.H., 2012. Developing a Greenhouse Gas Emissions Accounting Tool for Water Utilities. Master’s thesis. Clemson University, Anderson, S.C.

The authors thank those who participated in the Clemson survey or assisted in testing the GHG emissions accounting tool. They also thank Professor James Rieck of Clemson University’s Department of Applied Economics and Statistics for his assistance with development of the energy-use-prediction equations and statistical analysis.

Malcolm Pirnie, 2008. Statewide Assessment of Energy Use by the Municipal Water and Wastewater Sector. New York State Energy Research and Development Authority, Albany, N.Y.

About the authors

Tripathi, M., 2007. Life-Cycle Energy and Emissions for Municipal Water and Wastewater Services: Case-Studies of Treatment Plants in US. Master’s thesis. University of Michigan, Ann Arbor, Mich.

Anthony H. Johnston is a process engineer with Mustang Engineering, Greenville, S.C. He holds an MSc degree in environmental engineering and sciences and an MBA, both from Clemson University. Tanju Karanfil (to whom correspondence should be addressed) is a professor and chair of the Department of Environmental Engineering and Earth Sciences at Clemson University, 342 Computer Court, Anderson, SC 29625; [email protected].

Peer Review Date of submission: 10/28/2012 Date of acceptance: 04/04/2013

Raluy, R.G.; Serra, L.; Uche, J.; & Valero, A., 2005. Life Cycle Assessment of Water Production Technologies—Part 2: Reverse Osmosis Desalination Versus the Ebro River Water Transfer. International Journal of Life Cycle Assessment, 10:5:346. Stokes, J. & Horvath, A., 2006. Life Cycle Energy Assessment of Alternative Water Supply Systems. International Journal of Life Cycle Assessment, 11:5:335.

USEPA (US Environmental Protection Agency), 2012. Greenhouse Gas Reporting Program. www.epa.gov/ghgreporting (accessed July 20, 2011). USEPA, 2010. eGRID (Emissions & Generation Resource Integrated Database). www.epa.gov/cleanenergy/energy-resources/egrid/index.html (accessed July 20, 2011). Veerapaneni, S.; Klayman, B.; Wang, S.; & Bond, R., 2011. Desalination Facility Design and Operation for Maximum Efficiency. Water Research Foundation, Denver. Vince, F.; Aoustin, E.; Bréant, P.; & Marechal, F., 2008. LCA Tool for the Environmental Evaluation of Potable Water Production. Desalination, 220:1–3:37. Wallis, M.J.; Ambrose, M.R.; & Chan, C.C., 2008. Climate Change: Charting a Water Course in an Uncertain Future. Journal AWWA, 100:6:70. WRI (World Resources Institute) & World Business Council for Sustainable Development, 2004. The Greenhouse Gas Protocol: A Corporate Accounting and Reporting Standard. WRI, Washington.

2013 © American Water Works Association