EPA 600/R-14/376 I October 2014 I www.epa.gov/research
United States
Environmental Protection
Agency
Environmental and Cost Life Cycle
Assessment of Disinfection Options
for Municipal Drinking Water
Treatment
Office of Research and Development
National Homeland Security Research Center
-------
ENVIRONMENTAL AND COST LIFE CYCLE
ASSESSMENT OF DISINFECTION OPTIONS FOR
MUNICIPAL DRINKING WATER TREATMENT
Sarah Cashman2, Anthony Gaglione2, Janet Mosley2, Lori Weiss2, Troy R.
Hawkins1, Nick J. Ashbolt3, Jennifer Cashdollar3, Xiaobo Xue4, Cissy Ma1, and
Sam Arden5
National Risk Management Research Laboratory, U.S. Environmental Protection
Agency Office of Research and Development
2Eastern Research Group, Inc. (ERG)
3National Exposure Research Laboratory, U.S. Environmental Protection Agency,
Office of Research and Development
4Oak Ridge Institute for Science and Education (ORISE) Postdoctoral Research
Participant
5 University of Florida Engineering School of Sustainable Infrastructure &
Environment
Date: October 13, 2014
-------
ABSTRACT
EPA is evaluating water disinfection technologies in coordination with the Confluence Water
Technology Innovation Cluster (WTIC) and EPA's National Risk Management Research
Laboratory (NRMRL). EPA developed an environmental life cycle assessment (LCA) and cost
analysis to evaluate the environmental outcomes and costs associated with innovative
disinfection water treatment technologies. EPA is also interested in establishing an LCA and cost
model framework that could be used to study other technologies or changes to drinking water
and municipal wastewater treatment systems in the future. For each technology, there are
associated differences in pathogen removal, disinfection by-product formation, treatment facility
energy use and operating costs, input chemical requirements, and supply chain impacts.
This document summarizes the data collection, analysis, and results for a base case drinking
water treatment (DWT) plant reference model and alternative disinfection technologies. The base
case is modeled after the Greater Cincinnati Water Works (GCWW) Richard Miller Treatment
Plant. The infrastructure and operational datasets collected through iterative inquires and onsite
visit were used to develop the baseline life cycle model for the drinking water treatment system.
Results of the base case analysis show global warming, energy demand, fossil depletion,
acidification, human health cancer, human health criteria, and ecotoxicity impacts are largely
driven by electricity consumption at the drinking water treatment plant and during distribution to
the consumer. Labor and energy costs are the largest contributions to DWT plant costs. Disposal
of sedimentation waste is the greatest contributor to eutrophication. Source water acquisition
accounts for the majority of blue water use, with 1.2 m3 of source water from the river required
to deliver 1 m3 of water to the consumer. Metal depletion impacts are primarily governed by
chemical usage in the pre-disinfection and fluoridation stages as well as infrastructure
requirements at the DWT plant and distribution network. Overall, the primary disinfection with
gaseous chlorine life cycle stage only contributes zero to five percent to the total life cycle
impacts of DWT for the results categories examined. LCA and cost results decrease slightly
when excluding the adsorption step (0-15 percent).
EPA compared the base case results to four in-plant disinfection alternatives. The disinfection
alternatives considered are in different stages of development. In-plant alternatives include
disinfection by ultraviolet (UV) light (conventional mercury-vapor bulb system, LED UV, and
plasma-bead UV) and oxidation/disinfection using ferrate ions. The in-plant alternatives would
reduce the amount of chlorine required by the drinking water treatment plant among other
benefits. The datasets for compiling the life cycle inventory of disinfection technologies were
based on available industrial specifics and literature sources. Utilization of ferrate results in
environmental, human health, and cost benefits for combined use in the pre-disinfection and
primary disinfection stages, since ferrate acts as both a coagulant and disinfectant and only small
dosages are required for treatment. Application of UV technology increases impacts during
disinfection through increased electricity consumption and through new capital investment, but
eliminates the formation of disinfection by-products and greatly reduces hazardous chlorine
usage. LED UV is more energy efficient compared to conventional mercury-vapor UV; however,
it is currently developed only for point-of-use applications, and not large-scale treatment
facilities.
11
-------
In addition, EPA considered point-of-use disinfection alternatives such as disposable membrane
tap filters and small scale LED UV disinfection, which may be used in hospitals and other health
care facilities to further reduce exposure to pathogens for immune-compromised individuals.
Point-of-use disinfection alternatives for use at home were not considered. The point-of-use
technologies are add-on technologies and are not compared to the base case results. For hospital
point-of-use disinfection, the LED UV technology has the greater impacts overall compared to
the disposable membrane tap filter. The LED UV system requires 0.0039 kWh per m3 water
treated for operation; whereas, the disposable membrane tap filter does not require electricity for
generation.
In general, this analysis is provided to understand the potential impacts and trade-offs between
different drinking water disinfection technologies within the framework of the entire drinking
water supply system, and it is not intended to provide a recommendation on whether any
technology is superior to other technologies. The open-source and process based models built in
this study are flexible to incorporate future development of disinfection technologies and
associated datasets.
in
-------
Although the information in this document has been funded by the United States Environmental
Protection Agency under Contract EP-C-12-021 to Eastern Research Group, Inc., it does not
necessarily reflect the views of the Agency and no official endorsement should be inferred.
IV
-------
TABLE OF CONTENTS
1. INTRODUCTION AND STUDY GOAL 1
2. SCOPE 3
2.1 Functional Unit 3
2.2 System Boundaries 3
2.3 Impacts and Flows Tracked 5
2.3.1 Normalized and Weighted Results 7
3. LCA METHODOLOGY 8
3.1 Data Collection and Model 8
3.2 Unit Processes 9
3.3 Base Case Data Sources 10
3.4 Infrastructure Modeling 15
3.5 LCA Limitations 21
4. BASE CASE COST ANALYSIS 22
4.1 Base Case Data Sources 22
4.1.1 Annual Operating Costs 22
4.1.2 Capital Costs 23
4.2 Base Case Cost Method 24
4.3 Cost Data Quality, Assumptions, and Limitations 25
5. BASE CASE RESULTS 25
5.1 Base Case Normalized Results 31
5.2 Infrastructure Contribution to Base Case Results 31
6. BASELINE SENSITIVITY ANALYSES 34
7. IN-PLANT ALTERNATIVE DISINFECTION TECHNOLOGIES 42
7.1 System Boundaries 43
7.2 Aquionics Conventional UV Disinfection System 44
7.2.1 LCA Model 45
7.2.2 Cost Analysis 47
7.2.3 Results 48
7.3 Aquionics LED UV Disinfection System 53
7.3.1 LCA Model 53
7.3.2 Unit Processes 54
7.3.3 Results 55
7.4 Ferrate Technology 60
7.4.1 Data Collection and System Boundaries 60
7.4.2 LCA Model 62
7.4.3 Cost Analysis 64
7.4.4 Results 66
7.5 Imaging Systems Technology 74
7.6 Comparative Results 74
8. POINT-OF-USE ALTERNATIVE DISINFECTION TECHNOLOGIES 78
8.1 System Boundaries 78
8.2 Pall Point-of-Use Filter 79
8.2.1 LCA Model 79
8.2.2 Unit Processes 80
8.2.3 Results 81
8.3 LED UV Point-of-Use Filter 84
8.3.1 LCA Model 84
8.3.2 Unit Processes 85
8.3.3 Results 86
8.4 Comparative Results 88
-------
9. OVERALL RESULTS SUMMARY 91
10. REFERENCES 91
VI
-------
List of Tables
Table 1. Primary unit process matrix for the two base case models 5
Table 2. Impact and flow results categories 7
Table 3. Data sources 11
Table 4. Incoming and outgoing water quality metrics for GCWW Richard Miller
Treatment Plant (per m3water) 12
Table 5. Base case DWT LCI model-input and output operational data (per m3 drinking
water delivered to consumer) 13
Table 6. Infrastructure requirements for drinking water treatment plant buildings and
features (perm3 water delivered to consumer) 17
Table 7. Infrastructure requirements for drinking water treatment plant on-site piping (per
m3 water delivered to consumer) 18
Table 8. Infrastructure requirements for drinking water treatment distribution system
piping (perm3 water delivered to consumer) 19
Table 9. Infrastructure requirements for drinking water treatment distribution system
water storage, motors, pumps, and valves (per m3 water delivered to consumer) 20
Table 10. Annual Costs collected from GCWW 23
Table 11. GCWW Capital Improvement Projects Spending for Facilities and Water
Mains from 2000 to 2011 24
Table 12. Base Case 1 and Base Case 2 results per m3 drinking water delivered to the
consumer 30
Table 13. Contribution of infrastructure to Base Case 1 results per m3 drinking water
delivered to the consumer 33
Table 14. Sensitivity analyses for base case model runs 34
Table 15. U.S. electrical grid fuel profiles 35
Table 16. Unit process matrix for alternative disinfection technologies 43
Table 17. Conventional UV data sources 47
Table 18. Cost data provided by Aquionics.a 48
Table 19. Conventional UV results per m3 drinking water delivered to the consumer 49
Table 20. LED UV data sources 55
Table 21. LED UV results perm3 drinking water delivered to the consumer 56
Table 22. Ferrate data sources 64
Table 23. Cost data provided by FTT 64
Table 24. Ferrate results per m drinking water delivered to the consumer 69
Table 25. Point-of-Use hospital filter data sources 81
Table 26. Base case and Base case plus point-of-use hospital filter results per m3 drinking
water delivered to the consumer 82
Table 27. Point-of-Use LED UV data sources 86
Table 28. Base case and Base case plus point-of-use hospital LED UV disinfection results
perm3 drinking water delivered to the consumer 87
vn
-------
List of Figures
Figure 1. System boundaries of drinking water treatment base case options 4
Figure 2. System boundaries of drinking water treatment base case showing infrastructure
input 16
Figure 3. Base Case 1 contribution analysis results 27
Figure 4. Base Case 1 contribution analysis results with unit process detail 28
Figure 5. Base Case 1 and Base Case 2 comparative summary results 29
Figure 6. Percent change in impacts if adsorption is excluded 30
Figure 7. Base case normalized results 31
Figure 8. Infrastructure contribution analysis 32
Figure 9. Significance of electricity mix: RFC West versus U.S. average baseline 37
Figure 10. Tornado chart of the sensitivity results for the relative changes in total costs
for Base Case 1 37
Figure 11. Base Case 1 DWTP electricity usage sensitivity analysis 38
Figure 12. Base case 1 distribution system electricity usage sensitivity analysis 38
Figure 13. Base Case 1 chlorine usage sensitivity analysis 39
Figure 14. Base Case 1 DWTP infrastructure lifetime sensitivity analysis 39
Figure 15. Base Case 1 distribution system infrastructure lifetime sensitivity analysis 40
Figure 16. Base Case 1 alum coagulant usage sensitivity analysis 40
Figure 17. Base Case 1 lime usage sensitivity analysis 41
Figure 18. Base Case 1 distribution sodium hypochlorite usage sensitivity analysis 41
Figure 19. Base Case 1 natural gas for GAC reactivation sensitivity analysis 42
Figure 20. System boundaries of drinking water treatment base case and in-plant
disinfection alternatives 44
Figure 21. Base Case 1 and conventional UV comparative summary results 50
Figure 22. Percent change in impacts if using conventional UV rather than gaseous
chlorine for disinfection 51
Figure 23. Conventional UV electricity usage sensitivity analysis 52
Figure 24. Tornado chart of the sensitivity results for the relative changes in total costs
for the conventional UV scenario 53
Figure 25. Base Case 1 and LED UV comparative summary results 57
Figure 26. Percent change in impacts if using LED UV rather than gaseous chlorine for
disinfection 58
Figure 27. LED UV electricity usage sensitivity analysis 59
Figure 28. System boundaries of ferrate drinking water treatment 62
Figure 29. Base Case 1, Base Case 2, and ferrate comparative results by life cycle stage 70
Figure 30. Percent reduction when switching from Base Case 1 to ferrate DWT system 71
Figure 31. Ferrate dosage sensitivity analysis 73
Figure 32. Tornado chart of the sensitivity results for the relative changes in total costs
for the ferrate scenario 74
Figure 33. Summary comparative results of alternative disinfection technologies 76
Figure 34. Normalized comparative results for different drinking water treatment
disinfection technologies 77
Figure 3 5. System boundaries for hospital point-of-use drinking water treatment 79
Vlll
-------
Figure 36. Base Case 1 plus point-of-use hospital filter contribution analysis results 83
Figure 37. Base case percent change with point-of-use filter 84
Figure 38. Base Case 1 plus point-of-use hospital LED UV disinfection contribution
analysis results 88
Figure 39. Comparative results for different hospital point-of-use disinfection
technologies 90
IX
-------
1. INTRODUCTION AND STUDY GOAL
This study investigates disinfection technologies that are currently under development in the
Cincinnati Region in coordination with the Confluence Water Technology Innovation Cluster1
and EPA's National Risk Management Research Laboratory. Each technology provides an
alternative means of disinfecting drinking water and may address goals related to reducing
disinfection by-products, improving microorganism and virus reduction, reducing life-cycle
impacts, or reducing disinfection costs.
EPA collected data from the Greater Cincinnati Water Works (GCWW) Richard Miller
Treatment Plant to develop a base case drinking water treatment (DWT) plant LCA model and
cost analysis. The base case GCWW plant is a 106 million gallon per day (MGD) plant, which
uses gaseous chlorine as the primary disinfectant. GCWW uses a granular activated carbon
(GAC) system for removal of organics prior to chlorine addition. Additional details on the base
case plant are provided in Sections 2.2 and 3.2. This study evaluated base case models with and
without GAC. The goal for the base case LCA model and cost analysis is to:
1. Evaluate the base case environmental outcomes and costs to provide a baseline for
comparison to alternative disinfection technologies.
2. Establish an LCA and cost model framework that could be used to study other
technologies or changes to DWT systems.
9
This study addresses the following research questions :
1. What are the net life cycle impacts associated with drinking water treatment from source
water acquisition through distribution?
2. What are the contributions of each life cycle stage to the net result for each impact
category?
3. How do the two different base-case drinking water treatment options compare to one
another for each impact category?
4. How do the impacts and costs change as parameters associated with disinfection, energy
use, and disinfection by-products (DBF) vary? What parameters associated with
electricity use have the greatest effect on impacts and costs?
This study compared the results of the base case analysis to four in-plant disinfection alternatives
and examined the additional impact of applying two point-of-use disinfection technologies for
hospitals:
Confluence is a network of water technology researchers, businesses, utilities, and others in the southwest Ohio,
northern Kentucky, and southeast Indiana region. The group was formed in 2011 with help from EPA and the U.S.
Small Business Administration. See http://www.watercluster.org and http://www2.epa.gov/clusters-program for
more information.
2 This project requires the collection and use of existing data. EPA developed a Quality Assurance Project Plan
(QAPP) which outlines the quality objectives for this project. The plan is entitled Quality Assurance Project Plan
for Systems-Based Sustainability and Emerging Risks Performance Assessment of Cincinnati Regional Water
Technology Innovations: Comparative Life Cycle Assessment and Cost Analysis of Water Treatment Options, and
was prepared by Eastern Research Group, Inc. for U.S. EPA Sustainable Technology Division, National Risk
Management Research Laboratory. The plan was approved February 2013.
1
-------
• Conventional ultraviolet (UV) disinfection - UV radiation can effectively treat drinking
water for viruses and bacteria. This study evaluated replacing chlorine disinfection with a
conventional UV system. EPA worked with Aquionics to develop the conventional UV
disinfection model. Recently GCWW installed a UV system to use in conjunction with
traditional chlorine disinfection to improve removal of pathogens.
• LED UV disinfection - Use of a large-scale LED UV system. EPA worked with
Aquionics to develop the LED UV disinfection model.
• Plasma-bead UV disinfection - Use of a new technology developed by Imaging Systems,
which generates UV light for disinfection.
r\
• Ferrate disinfection - Use of ferrate (FeC>4 "), a strong oxidizer, for disinfection. Ferrate
Treatment Technologies, LLC (FTT) has developed an on-site, skid-mounted method of
producing ferrate for disinfection. Ferrate can also be used during pre-treatment as a
coagulant.
• Point-of-use disinfection using disposable membrane tap filters and small scale LED UV
disinfection.
Each technology has differences in pathogen removal, disinfection by-product formation,
chemical and energy requirements, costs, and environmental benefits. EPA intends to answer the
following research questions through the disinfection alternative analysis:
1. What are the net life cycle impacts associated with each disinfection alternative (in-plant
and point-of-use add-on)?
2. For which life cycle stages do the results for the in-plant alternatives differ from the base
case?
3. How do the overall plant costs change for each of the in-plant alternatives?
4. What are the costs and environmental impacts of additional reductions in pathogens for
point-of-use add-ons in a health care facility?
The remainder of the report provides details on the analysis and is organized into the following
sections:
• Section 2 defines the study scope.
• Section 3 provides details on the LCA methodology including a description of the unit
processes included in the base case model.
• Section 4 describes the cost analysis.
• Section 5 presents base case results.
• Section 6 presents base case sensitivity results.
• Section 7 describes the in-plant alternative disinfection technologies, modifications to the
LCA model and cost analysis, and results for each technology.
• Section 8 describes the point-of-use alternatives and compares costs of the alternative to
additional pathogen reduction.
• Section 9 summarizes the study results.
• Section 10 provides the references for the study.
-------
2. SCOPE
The base case DWT model includes source water acquisition, pre-disinfection, primary
disinfection, and distribution. This study examined environmental impacts and changes in costs
for different disinfection technologies; therefore, the base case established the reference case for
comparison to alternative drinking water disinfection technologies.
2.1 Functional Unit
The functional unit, which provides the basis for comparison, used in this study is the delivery to
the consumer of one cubic meter of water that meets or exceeds National Primary Drinking
Water Regulations for microorganisms, disinfectants, disinfectant by-products, inorganics,
organics, and radionuclides.3 For the point-of-use technology analysis, the drinking water
delivered to the consumer has a greater reduction in pathogens compared to the base case and in-
plant disinfection technology alternatives analysis. Results for the point-of-use analysis are,
therefore, not compared directly to results for the base case or in-plant alternative technologies as
the end product delivered by these different pathways are not functionally equivalent.
2.2 System Boundaries
Figure 1 illustrates the system for the DWT base case model. The system boundaries start at
acquisition of source water from a river and end at delivery of the treated drinking water to the
consumer. Transportation requirements for all inputs to the processes within supply chains, such
as transporting alum coagulant to the treatment plant, are also included as are all capital
equipment and infrastructure requirements for the drinking water treatment plant and distribution
network. Impacts for the following two base case model runs were evaluated:
• Base Case 1: Representative moderate-sized water treatment facility, including GAC
adsorption.
• Base Case 2: Variation of the base case excluding GAC adsorption.
3 U.S. EPA (2013) "National Primary Drinking Water Regulations"
http ://water. epa.gov/drink/contaniinants/index. cfm
-------
Source Water, River j
V . ^y
Source Water
Acquisition
Flocculation
Sand Production
Lime Production
Sedimentation
5
Filtration
1
«
1
Iron Sulfate
Production
Disposal of
Sedimentation
Waste
w/
adsorption
w/o
adsorption
Sodium
Hexametaphosphate
Production
Transport, Treated
Drinking Water, Water
Supply Pipeline
LEGEND
Primary Input/
Final Demand
Primary Process
Reference
Supply Chain
Or gate (multiple
outputs)
Or gate (multiple
inputs)
Figure I. System boundaries of drinking water treatment base case options.
Drinking water treatment operations along with infrastructure raw material extraction and
construction are within the system boundaries. End-of-life of infrastructure is excluded
due to lack of available data.
The GCWW Richard Miller Treatment Plant serves customers in Cincinnati and the surrounding
towns in Ohio and Kentucky. The plant has 241,000 customers, of which 230,000 are residential
customers. The remaining are either businesses or other non-residential customers. In terms of
-------
volume of water supplied, 16 percent of the treated water volume goes to wholesale customers,
48 percent of the treated water volume goes to residential customers, and 36 percent of the
treated water volume goes to non-residential customers. GCWW was the first major municipal
water provider to utilize granular activated carbon for adsorption of toxins.4 While this
adsorption process offers health benefits for consumers of GCWW water, it is not used in other
areas and is not necessary for some other water sources. While GCWW was used as a reference
for this study to allow the results to be based on actual operating conditions, the intent is that the
results can be extrapolated to provide input to decisions made for other systems as well. For this
reason, a second set of base case results without the GAC adsorption process are provided.
Table 1
shows the primary life cycle stages and unit processes included in Base Case 1 and 2
Table 1. Primary unit process matrix for the two base case models.
Life Cycle Stage
Reported
Source Water Acquisition
Drinking Water Treatment
Plant, Energy and
Infrastructure
Pre-Disinfection
Primary Disinfection
Distribution
Use
Unit Processes Covered
Source Water Acquisition
Drinking Water Treatment Plant,
Energy Usage
Flocculation
Alum Coagulant
Sedimentation
Disposal, Sedimentation Waste
Filtration
Adsorption
GAC Production
GAC Regeneration
Primary Disinfection, Gaseous
Chlorine
Fluoridation
Transport, Treated Drinking
Water, Water Supply Pipeline
Distribution Infrastructure,
Drinking Water
Drinking Water Consumption
ctf
O
ctf
pq
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
(N
ctf
O
ctf
pq
X
X
X
X
X
X
X
X
X
X
X
X
2.3 Impacts and Flows Tracked
The full inventory of emissions generated in an LCA study is lengthy and diverse, making it
difficult to interpret emissions profiles in a concise and meaningful manner. Life Cycle Impact
Activated Carbon: Solutions for Improving Water Quality, Zaid K. Chowdhury, Garret P. Westerhoff, R. Scott
Summers, Brian Leto, Kirk Nowack, American Water Works Association, 2012.
-------
Assessment (LCIA) helps with interpretation of the emissions inventory. In the LCIA phase, the
inventory of emissions is first classified into categories in which the emissions may contribute to
impacts on human health or the environment. Within each impact category, the emissions are
then normalized to a common reporting basis, using characterization factors that express the
impact of each substance relative to a reference substance.
Table 2 shows the complete list of impacts examined for the base case model runs. This study
addresses global, regional, and local impact categories. The LCIA method provided by the Tool
for the Reduction and Assessment of Chemical and Environmental Impacts (TRACI), version
2.0, developed by the U.S. EPA specifically to model environmental and human health impacts
in the U.S., is the primary LCIA method applied in this work.5 Additionally, the ReCiPe LCIA
method is used to characterize fossil fuel, blue water use (i.e. water depletion), and metal
depletion.6 Energy is tracked based on point of extraction using the cumulative energy demand
method developed by Ecoinvent.7 The blue water use impact category represents freshwater use
from surface water or groundwater sources. The blue water use category includes indirect
consumption of water from upstream processes, such as water withdrawals for electricity
generation (e.g., evaporative water losses from coal power cooling water and establishment of
o
hydroelectric dams). Some flows specific to drinking water treatment, and not typically reported
in LCA studies, are included in the results reported in the analysis:
• Cryptosporidium (Crypto) Exposure
• Total Trihalomethanes (TTHM) Exposure
• Chlorine Usage
These unique flows are tracked based on data reported by GCWW for specific life cycle stages,
and do not cover all potential upstream exposure to cryptosporidium and TTHM or upstream use
of chlorine. The purpose of tracking and displaying these aspects is to provide a more balanced,
albeit cursory, analysis of other benefits associated with the disinfection technologies addressed
by this study. These results are intended to be supplemented by additional studies focused on
providing better resolution of these aspects for decision-making purposes within the context of a
specific system. They are provided here for context.
5 EPA's Tool for the Reduction and Assessment of Chemical and Other Environmental Impacts (TRACI), see:
http://www.epa.gov/nrmrl/std/sab/traci/.
6 Goedkoop M.J., Heijungs R, Huijbregts M, De Schryver A.; Struijs I, Van Zelm R, ReCiPe 2008, A life cycle
impact assessment method which comprises harmonised category indicators at the midpoint and the endpoint level;
First edition Report I: Characterisation; 6 January 2009, http://www.lcia-recipe.net
7 Ecoinvent Cumulative Energy Demand (CED) Method implemented in ecoinvent data v2.2. 2010. Swiss Centre
for Life Cycle Inventories.
8Pfister, S., Saner, D., Koehler, A. 2011. The environmental relevance of freshwater consumption in global power
production. International Journal of Life Cycle Assessment, 16 (6): 580-591.
-------
Table 2. Impact and flow results categories.
Category
Cost
Crypto Exposure
TTHM Exposure
Chlorine Usage
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health,
Cancer, Total
Human Health,
NonCancer, Total
Human Health,
Criteria
Ecotoxicity, Total
Methodology
Cost Analysis
Individual Flow
Individual Flow
Individual Flow
TRACI 2.0
ecoinvent
ReCiPe
TRACI 2.0
TRACI 2.0
Custom
TRACI 2.0
TRACI 2.0
ReCiPe
TRACI 2.0
TRACI 2.0
TRACI 2.0
TRACI 2.0
Unit
$
oocyst
kg TTHM
kgd2
kg CO2 eq
MJeq
kg oil eq
H+ moles eq
kgNeq
m3
kg O3 eq
kgCFC-lleq
kg Fe eq
CTU
CTU
kg PM10 eq
CTU
Description
Measures total cost in U.S. dollars.
Measures exposure of consumer to cryptosporidium in
delivered drinking water. Cryptosporidum levels reported
in Distribution life cycle stage.
Measures exposure of consumer to TTHM in delivered
drinking water. TTHM levels reported in Distribution life
cycle stage.
Measures gaseous chlorine usage for primary
disinfection, which indicates on-site storage of this
hazardous chemical.
Represents the potential heat trapping capacity of
greenhouse gases.
Measures the total energy use from point of extraction.
Assesses the potential reduction of fossil fuel energy
resources.
Quantifies the potential acidifying effect of substances on
their environment.
Assesses potential impacts from excessive load of macro-
nutrients to the environment.
Calculates consumptive use of fresh surface or
groundwater.
Determines the potential formation of reactive substances
(e.g. tropospheric ozone) that cause harm to human health
and vegetation.
Measures potential stratospheric ozone depletion.
Assesses the potential reduction of metal resources.
A comparative toxic unit (CTU) for cancer characterizes
the probable increase in cancer related morbidity (from
inhalation or ingestion) for the total human population
per unit mass of a chemical emitted.
A CTU for noncancer characterizes the probable increase
in noncancer related morbidity (from inhalation or
ingestion) for the total human population per unit mass of
a chemical emitted.
Assesses human exposure to elevated paniculate matter
less than 10 um.
Assesses potential fate, exposure, and effect of chemicals
on the environment.
2.3.1 Normalized and Weighted Results
Normalization is an optional step in LCA that aids in understanding the significance of the
impact assessment results. Normalization is conducted by dividing the impact category results by
a normalized value. The normalized value is typically the environmental burdens of the region of
interest either on an absolute or per capita basis. The results presented here are normalized to
reflect person equivalents in the U.S. using TRACI v2.1 normalization factors.9 Only impacts
with TRACI normalization factors are shown. Some categories like blue water use and energy
demand are excluded due to lack of available normalization factors.
Ryberg, M., Vieira, M.D.M., Zgola, M., Bare, I, and Rosenbaum, R.K., 2014. Updated US and Canadian
normalization factors for TRACI 2.1. Clean Techn Environ Policy, 16: 329-339.
7
-------
Weighting is an additional optional step in LCA that provides a link between the quantitative
results and subjective choices of decision makers. This study applies weights to the normalized
results described above. The weights utilized here were developed by the National Institute of
Standards and Technology (NIST) for the BEES (Building for Environmental and Economic
Sustainability) software.10 This weighting set was created specifically for the buildings sector
context, which may not be completely compatible with the water treatment sector. However, due
to lack of a weighting set specific to the water treatment sector, this NIST weighting set has been
utilized.
3. LCA METHODOLOGY
Development of a life cycle assessment requires significant input data, an LCA modeling
platform, and impact assessment methods. This section provides background on the development
of the LCA model. Section 3.1 discusses the data collection method and model, Section 3.2
describes the unit processes, Section 3.3 lists the data sources, and Section 3.4 describes
limitations of the LCA model.
In this study, GCWW provided much of the LCA input data for the unit processes listed in Table
1. This study also used publicly accessible and private databases to provide underlying data sets
describing the supply chains of inputs to the processes modeled here. For example, in addition to
the unit processes described in Section 3.2, an LCA also includes impacts from the production of
any materials required in the process.
3.1 Data Collection and Model
The accuracy of the study is directly related to the quality of input data. Data were collected
electronically using Excel templates designed by the project team to be completed by GCWW.
Data collection was an iterative process whereby the project team asked GCWW multiple rounds
of questions to ensure all necessary life cycle and cost information was being reported and
properly interpreted in the assessment. The quality and objectivity of results were ensured
through carefully adhering to the data collection protocols and quality procedures laid out in the
Quality Assurance Project Plan prior to beginning work on the project.
Each unit process in the life cycle inventory was constructed independently of all other unit
processes. This allows objective review of individual data sets before their contribution to the
overall life cycle results has been determined. Also, because these data are reviewed
individually, EPA reviewed assumptions based on their relevance to the process rather than their
effect on the overall outcome of the study.
The model was constructed in OpenLCA, an open-source LCA software package provided by
GreenDelta.
10 Gloria, T.P., Lippiatt, B.C., and Cooper, J. 2007. Life cycle impact assessment weights to support environmentally
preferable purchasing in the United States. Environ. Sci. Technol, 41, 7551-7557.
-------
3.2 Unit Processes
EPA developed new unit processes for the specific DWT processes listed below (categorized by
the overall life cycle stage). As shown in Figure 1, the DWT base case unit processes start
with source water acquisition and end with drinking water use. Unit processes from background
LCI database (e.g., ecoinvent v2.2 and U.S. LCI) that have not been modified are identified in
Section 3.3, Table 3 (Data Sources). On-site DWT plant infrastructure is included for each unit
where applicable. Section 3.4 covers the details of the infrastructure modeling.
Drinking water acquisition
1. Source Water Acquisition. The GCWW Miller Plant used for the base case model uses
surface water from a river for the raw water source.
Drinking Water Treatment Plant, Energy Usage
2. Drinking Water Treatment Plant, Energy Usage. Covers all electricity required to
pump in raw water, and pumping energy throughout the drinking water treatment plant.
Pre-Disinfection
3. Flocculation. Aggregates suspended solids by adding coagulant and coagulant aid and
mixing to increase the particle size to allow settling. Alum is modeled as the coagulant.
4. Alum Coagulant. Production of average alum derived from industrial aluminum sulfate.
5. Sedimentation. Removes suspended solids from water by gravity settling. In an
intermediate step "lime addition", lime is added in clarification basins prior to filtration.
6. Disposal, Sedimentation Waste. Disposal of settled solids to surface water.
7. Filtration. Removes remaining solids from water using a sand filter. Includes
replacement of filter materials during normal operation life of the filter.
8. Adsorption. Removes organics by a granular activated carbon (GAC) system. As noted
in Table 1, adsorption is not included in Base Case 2.
9. Granular Activated Carbon Production. Production of average U.S. granular activated
carbon from bituminous coal. GAC production is not included in Base Case 2.
10. Granular Activated Carbon Regeneration. Regenerates activated carbon by
conventional thermal regeneration method with natural gas as the required energy source.
Also includes carbon loss and replacement. GAC regeneration is not included in Base
Case 2.
11. Conditioning. Adjust pH using sodium hydroxide and addition of a polyphosphate,
sodium hexametaphosphate.
12. Pre-Disinfection. This unit process aggregates the upstream pre-disinfection unit
processes from flocculation through conditioning.
Disinfection
13. Primary Disinfection, Gaseous Chlorine. Representative of a conventional DWT
system using gaseous chlorine for primary disinfection.
Distribution
14. Fluoridation. Hydrofluorosilicic acid addition prior to distribution.
15. Transport, Treated Drinking Water, Water Supply Pipeline. Transporting treated
drinking water to end users. Accounts for pumping energy.
-------
16. Valves for Distribution System. Steel required for production of valves for distribution
system.
17. Pumps for Distribution System. Cast iron and steel for production of pumps for
distribution system.
18. Motors for Distribution System. Steel, copper, aluminum, and cast iron for production
of motors for distribution system.
19. Water Storage Infrastructure. Concrete and steel for construction of water storage
tanks, earthworks associated with reservoir construction.
20. Distribution Pipe Network. Production and installation of concrete and iron pipes for
distribution system.
21. Distribution. This unit process aggregates upstream distribution unit processes including
fluoridation, pipeline transport of the treated drinking water, and infrastructure
components of the distribution system. Sodium hypochlorite is also added as an input to
the distribution life cycle stage as it is used in small amounts to boost the chlorine levels
in certain sections of the distribution system.
Use
22. Drinking Water Consumption. Final delivery of water to an average consumer. This
unit process aggregates the other main life cycle stages and is used to build the final
product system. There are no actual impacts associated with the drinking water
consumption life cycle stage itself.
3.3 Base Case Data Sources
Table 3 displays the data sources used for the DWT base case model. In general, data from
GCWW were used where available. GCWW provided data for their Richard Miller Treatment
Plant, which produces approximately 106 MGD of finished drinking water. The incoming and
outgoing water quality metrics for the Richard Miler Treatment Plant reported for this study are
shown in Table 4. For upstream processes that would not be known by GCWW such as
information on chemical production, EPA used information from the National Renewable
Energy Laboratory's U.S. Life Cycle Inventory Database (U.S. LCI), a publically available life
cycle inventory source.11 Where data were not available from GWCC or the U.S. LCI, EPA used
ecoinvent v2.2, a private Swiss LCI database with data for many unit processes.12 Table 5
presents the complete DWT base case LCI data used in the model on the basis of one cubic meter
of drinking water delivered to the consumer.
11 National Renewable Energy Lab. US LCI Database. See: http://www.nrel.gov/lci/database/default.asp.
12 Ecoinvent Centre (2010), ecoinvent data v2.2. ecoinvent reports No. 1-25, Swiss Centre for Life Cycle
Inventories.
10
-------
Table 3. Data sources.
Process
Source Water Acquisition
Incoming Transport of Chemicals to DWTP
Gaseous Chlorine Production
Sodium Chloride Production
Flocculation
Aluminum Sulfate Production (Powder)
Sulfuric Acid Production
Aluminum Hydroxide Production
Iron Sulfate
Sedimentation (Operation)
Disposal of Sedimentation Waste
Filtration (Operation)
Sand Production (for Use in Filter)
Adsorption
GAC Production
GAC Regeneration
Bituminous Coal Production
Conditioning
Sodium Hydroxide Production
Sodium Hexametaphosphatea
Lime Production
Primary Disinfection
Gaseous Chlorine Production
Sodium Hypochlorite Production
Fluoridation
Hydrofluorosilicic Acid Production11
Distribution (Operation)
Background Fuels and Energy
Infrastructure at the DWT Plant
Infrastructure in the Distribution System
Background Transportation Processes
Data Source
Data Collection
Data Collection
ecoinvent v2.2
ecoinvent v2.2
Data Collection
ecoinvent v2.2
ecoinvent v2.2
ecoinvent v2.2
ecoinvent v2.2
Data Collection
Data Collection
Data Collection
ecoinvent v2.2
Data Collection
Data Collection
Data Collection
U.S. LCI
Data Collection
ecoinvent v2.2
ecoinvent v2.2
U.S. LCI
Data Collection
ecoinvent v2.2
ecoinvent v2.2
Data Collection
ecoinvent v2.2
Data Collection
U.S. LCI
Data Collection
Data Collection
U.S. LCI
•GCWW
•GCWW
-GCWW
•GCWW
•GCWW
•GCWW
•GCWW
•GCWW
•GCWW
-GCWW
-GCWW
•GCWW
•GCWW
•GCWW
•GCWW
a Using sodium tripolyphosphate as surrogate, since no available LCI data exists for sodium
hexametaphosphate.
b Using hydrogen fluoride (HF) as surrogate, since fluorosilicic acid is a by-product of HF production, and no
available LCA data exists for hydrofluorosilicic acid production.
11
-------
Table 4. Incoming and outgoing water quality metrics for GCWW Richard Miller Treatment Plant
(per m3water).
Water Metrics
Ammonia
Arsenic
Chromium
Dissolved
organic carbon
Dissolved solids
Iron
Manganese
Nitrate
pH
Phosphorus
Suspended
solids
Temperature
Total organic
carbon
Turbidity
TTHM
Chlorine
Cryptosporidium
Giardia
E. coli
Heterotrophic
plate count
Incoming Water
Minimum
0.010
O.001
<5.0E-04
2.50
158
0.30
0.053
0.63
7.50
0.030
1.90
5.10
2.10
2.40
<5.0E-04
0
<20.0
<20.0
0
0
Maximum
0.19
0.0016
0.0029
3.30
299
0.30
0.053
1.14
8.40
0.11
225
33.0
4.80
307
<5.0E-04
0
<91.0
200
6,030,000
14,000,000,000
Average
0.050
O.001
0.0010
2.99
229
0.30
0.053
0.89
7.80
0.060
43.1
18.0
3.05
46.0
<5.0E-04
0
<51.0
20.0
1,340,000
2,250,000,000
Outgoing Water
Minimum
0
O.001
O.010
0.61
132
O.020
0.010
0.62
8.20
0.15
0
4.70
0.40
0.050
0.0078
1.13
0.80
O.80
0
0
Maximum
0
O.001
O.010
1.01
317
O.020
0.010
1.06
8.80
0.20
0
29.0
1.43
0.13
0.020
1.66
<1.10
<1.10
0
0
Average
0
O.001
O.010
0.94
228
O.020
0.010
0.86
8.80
0.17
0
17.0
0.85
0.070
0.016
1.37
<1.00
<1.00
0
0
Unit (per
m3 water)
g
g
g
g
g
g
g
g
pH
g
g
°C
g
NTU
g
g
oocysts
oocysts
counts
counts
Source: GCWW primary data collection for the year 2011.
12
-------
Table 5. Base case DWT LCI model-input and output operational data (per m3 drinking water delivered to consumer).
Unit
TOTAL
Quantity
Quantity by Life Cycle Stage
Source Water
Acquisition
Energy for
Pumping
Flocculation
Sedimentation
Disposal,
Sedimentation
Waste
Lime Addition
Filtration
Adsorption
GAC
Reactivation
Conditioning
d
|1§1
C ft
-------
Combination truck transport,
sodium hypochlorite
Combination truck transport,
sodium hydroxide
Barge transport, sodium hydroxide
Combination truck transport
sodium hexametaphosphate
Combination truck transport
polymer (polyDADMAC)
Combination truck transport GAC
Unit
tkm
tkm
tkm
tkm
tkm
tkm
TOTAL
Quantity
2.7E-05
2.7E-05
6.0E-04
1.3E-04
9.8E-04
6.5E-04
Quantity by Life Cycle Stage
Source Water
Acquisition
Energy for
Pumping
Flocculation
9.8E-04
Sedimentation
Disposal,
Sedimentation
Waste
Lime Addition
Filtration
Adsorption
6.5E-04
GAC
Reactivation
Conditioning
2.7E-05
6.0E-04
1.3E-04
£
|lll
e .5 w °
fl 5 o 0
Fluoridation
Distribution
2.7E-05
3
bfL =
% « s - g-^
s oj .S es ft q
5 H Q > £ '£.
Drinking
Water
Consumption
Outputs
Waste &
Loss
Water
Emissions
Air
Emissions
Final
Product
Disposal of sedimentation waste
Water loss
Aluminum (water emissions)
Ammonia (water emission)
Biological oxygen demand (water
emission)
Chemical oxygen demand (water
emission)
Suspended solids (water emission)
Carbon monoxide (air emission)
Nitrogen oxides (air emission)
Particulates, <10 um (air emission)
Particulates, <2.5 um (air
emission)
Sulfur oxides (air emission)
Volatile organic compounds (air
emission)
Drinking water delivered to
consumer
liters
m3
kg
kg
kg
kg
kg
kg
kg
kg
kg
kg
kg
m3
0.048
0.19
0.0016
3.6E-06
3.9E-04
0.0081
0.016
2.7E-05
9.2E-05
1.5E-05
1.5E-05
2.9E-04
1.3E-05
1.00
0.0036
0.048
0.0016
3.6E-06
3.9E-04
0.0081
0.016
2.7E-05
9.2E-05
1.5E-05
1.5E-05
2.9E-04
1.3E-05
0.19
1.00
Source: GCWW primary data collection from the year 2011.
14
-------
3.4 Infrastructure Modeling
Infrastructure at the drinking water treatment plant and for the distribution system was included
in the model based on primary data collected from GCWW. In the Figure 2 system boundaries,
infrastructure components modeled are shown in red. Each infrastructure component was
normalized to a cubic meter of water delivered to a consumer. It was assumed, based on
discussion with engineers at GCWW regarding replacement rates, that the lifetime of the
buildings, features, and pipes is 100 years. A shorter lifetime of 25 years was estimated for the
pumps and motors. Infrastructure was normalized by dividing the total infrastructure impact by
the total lifetime of the component, and then by the water delivered per year. It is assumed that
the water delivered per year (for every year during the infrastructure component lifetime) is
123,560,247 cubic meters, which is the volume of drinking water delivered to consumers in
2011. The infrastructure requirements for plant buildings and features (e.g., reservoirs, tanks), at
plant piping, distribution system piping, water storage in distribution system, and distribution
pumps and motors are shown in Table 6 through Table 9. To simplify the model for the
distribution system piping, only pipe types that represent more than 0.5 percent of the total
length were included. Pipe types greater than 0.5 percent of the length were then scaled up to
represent 100 percent of the total distribution system pipe length. Construction burdens were
determined based on the volume of earthworks required per infrastructure component.
15
-------
GCWW RICHARD MILLER TREATMENT PLANT
Source Water. Ri
", River
lntakeSl&S2
Pump Station
S1&S2
X
I
Source Water
Acquisition
\ Primary Input/
// Final Demand
Figure 2. System boundaries of drinking water treatment base case showing infrastructure input.
16
-------
Table 6. Infrastructure requirements for drinking water treatment plant buildings and features (per m3 water delivered to consumer).
Life Cycle Stage
Source Water
Acquisition
Flocculation
Sedimentation
Filtration
Adsorption
Conditioning
Primary Disinfection
Fluoridation
Infrastructure Component
Intake 1 (to Pump Station 1)
Intake 2 (to Pump Station 2)
Pump Station 1 (Near River)
Pump Station 2 (Farther from River)
Pretreatment Complex
Settling Reservoir #1 (Closer to
Pump Station)
Settling Reservoir #2 (Farther from
Pump Station)
Chemical House (East)
Clarification Basins
Filter Building
GAC Facility
Caustic Soda Facility
Chlorine Injector Facility
Clearwell #1
Clearwell #2
Material Type
Earthworks
(m3)
1.2E-05
1.2E-05
8.9E-07
3.3E-06
7.4E-06
4.8E-05
5.3E-05
9.3E-07
6.4E-06
5.2E-06
1.4E-05
4.1E-06
2.6E-07
9.9E-06
2.4E-06
Reinforcing
Steel (kg)
0
0
0
5.1E-05
1.1E-04
0
0
1.4E-05
0
8.0E-05
2.1E-04
6.4E-05
4.1E-06
0
0
6.5' Concrete
piping (m)
0
0
0
6.9E-08
1.5E-07
0
0
1.9E-08
0
1.1E-07
2.9E-07
8.6E-08
5.5E-09
0
0
Concrete
(m3)
0
0
0
6.0E-07
1.3E-06
0
0
1.7E-07
4.1E-07
9.3E-07
2.5E-06
7.4E-07
4.8E-08
4.7E-07
1.8E-07
Bricks
(kg)
9.4E-04
9.4E-04
0
0
0
1.8E-04
1.9E-04
0
0
0
0
0
0
0
0
Limestone
(kg)
0.0012
0.0012
0.0024
0
0
0
0
0
0
0
0
0
0
0
0
Source: GCWWprimary data collection with estimations made with facility map.
17
-------
Table 7. Infrastructure requirements for drinking water treatment plant on-site piping (per m3 water delivered to consumer).
Life Cycle
Stage
Source Water
Acquisition
Flocculation
Sedimentation
Filtration
Adsorption
Conditioning
Fluoridation
Diameter
7'
36"
50"
54"
72"
60"
72"
36"
54"
60"
72"
60"
36"
36"
36"
36"
Pipe Type
Gray Iron
Pipe (m)
0
2.7E-09
3.8E-08
1.9E-09
1.2E-09
6.5E-09
7.4E-08
2.3E-09
2.7E-09
7.3E-08
1.8E-08
3.7E-08
1.2E-08
6.9E-09
3.1E-09
1.5E-08
Ductile Iron
Pipe (m)
0
1.6E-09
2.3E-08
1.2E-09
7.0E-10
4.0E-09
4.5E-08
1.4E-09
1.6E-09
4.4E-08
LIE-OS
2.2E-08
7.4E-09
4.2E-09
1.9E-09
9.3E-09
Concrete
Pipe (m)
9.8E-08
2.4E-10
3.4E-09
1.7E-10
l.OE-10
5.9E-10
6.7E-09
2.1E-10
2.4E-10
6.6E-09
1.7E-09
3.3E-09
1.1E-09
6.2E-10
2.8E-10
1.4E-09
Total
Length
(m)
9.8E-08
4.6E-09
6.4E-08
3.3E-09
2.0E-09
LIE-OS
1.3E-07
3.9E-09
4.6E-09
1.2E-07
3.1E-08
6.3E-08
2.1E-08
1.2E-08
5.2E-09
2.6E-08
Earthworks
(m3)
l.OE-06
1.8E-08
3.5E-07
2.0E-08
1.7E-08
7.6E-08
1.1E-06
1.5E-08
2.8E-08
8.4E-07
2.7E-07
4.3E-07
7.0E-08
4.6E-08
2.1E-08
l.OE-07
Source: GCWW primary data collection with estimations made with facility map.
18
-------
Table 8. Infrastructure requirements for drinking water treatment distribution system piping (per m3 water delivered to consumer).
Life Cycle
Stage
Distribution
Diameter
0.75"
1"
1.5"
2"
2.5"
3"
4"
6"
8"
10"
12"
16"
20"
24"
30"
35"
36"
42"
44"
46"
48"
54"
60"
Pipe Type
Gray Iron
(m)
2.30E-09
6.76E-09
1.06E-08
2.90E-07
O.OOE+00
1.88E-08
1.52E-06
7.52E-05
9.28E-05
4.05E-06
3.21E-05
3.85E-06
8.10E-06
4.71E-06
9.96E-07
3.98E-06
4.55E-06
9.26E-07
2.95E-06
3.57E-07
1.29E-06
2.73E-07
3.06E-07
Ductile
Iron (m)
1.39E-09
4.10E-09
6.41E-09
1.76E-07
O.OOE+00
1.14E-08
9.19E-07
4.55E-05
5.62E-05
2.45E-06
1.94E-05
2.33E-06
4.91E-06
2.85E-06
6.03E-07
2.41E-06
2.76E-06
5.61E-07
1.79E-06
2.16E-07
7.82E-07
1.66E-07
1.85E-07
Concrete
(m)
2.07E-10
6.09E-10
9.52E-10
2.61E-08
O.OOE+00
1.69E-09
1.37E-07
6.77E-06
8.36E-06
3.65E-07
2.89E-06
3.47E-07
7.29E-07
4.24E-07
8.97E-08
3.58E-07
4.10E-07
8.34E-08
2.66E-07
3.22E-08
1.16E-07
2.46E-08
2.76E-08
Steel (m)
1.81E-11
5.32E-11
8.33E-11
2.28E-09
O.OOE+00
1.48E-10
1.19E-08
5.92E-07
7.31E-07
3.19E-08
2.53E-07
3.03E-08
6.38E-08
3.71E-08
7.84E-09
3.13E-08
3.58E-08
7.29E-09
2.32E-08
2.81E-09
1.02E-08
2.15E-09
2.41E-09
Copper
(m)
4.78E-12
1.41E-11
2.20E-11
6.03E-10
O.OOE+00
3.90E-11
3.15E-09
1.56E-07
1.93E-07
8.41E-09
6.67E-08
8.00E-09
1.68E-08
9.78E-09
2.07E-09
8.26E-09
9.45E-09
1.92E-09
6.13E-09
7.42E-10
2.68E-09
5.68E-10
6.36E-10
PVC (m)
1.26E-12
3.70E-12
5.78E-12
1.59E-10
O.OOE+00
1.03E-11
8.30E-10
4.11E-08
5.07E-08
2.21E-09
1.76E-08
2.11E-09
4.43E-09
2.57E-09
5.45E-10
2.17E-09
2.49E-09
5.06E-10
1.61E-09
1.95E-10
7.06E-10
1.49E-10
1.67E-10
HOPE
(m)
5.41E-12
1.59E-11
2.49E-11
6.82E-10
O.OOE+00
4.41E-11
3.57E-09
1.77E-07
2.18E-07
9.52E-09
7.55E-08
9.06E-09
1.90E-08
1.11E-08
2.34E-09
9.35E-09
1.07E-08
2.18E-09
6.94E-09
8.40E-10
3.04E-09
6.42E-10
7.19E-10
Transite
(m)
1.27E-11
3.73E-11
5.84E-11
1.60E-09
O.OOE+00
1.04E-10
8.38E-09
4.15E-07
5.12E-07
2.24E-08
1.77E-07
2.13E-08
4.47E-08
2.60E-08
5.50E-09
2.20E-08
2.51E-08
5.11E-09
1.63E-08
1.97E-09
7.13E-09
1.51E-09
1.69E-09
Total
Length
(m)
3.95E-09
1.16E-08
1.81E-08
4.97E-07
O.OOE+00
3.22E-08
2.60E-06
1.29E-04
1.59E-04
6.94E-06
5.50E-05
6.60E-06
1.39E-05
8.07E-06
1.71E-06
6.82E-06
7.80E-06
1.59E-06
5.06E-06
6.13E-07
2.21E-06
4.68E-07
5.25E-07
Earthworks
(m3)
4.21E-09
1.25E-08
2.01E-08
5.67E-07
O.OOE+00
3.87E-08
3.29E-06
1.80E-04
2.43E-04
1.16E-05
9.97E-05
1.40E-05
3.39E-05
2.25E-05
5.71E-06
2.62E-05
3.08E-05
7.30E-06
2.44E-05
3.10E-06
1.17E-05
2.83E-06
3.58E-06
Source: GCWW Primary data collection from 2011 water main inventory.
19
-------
Table 9. Infrastructure requirements for drinking water treatment distribution system water storage, motors, pumps, and valves (per m3
water delivered to consumer).
Life Cycle
Stage
Distribution
Infrastructure
Water Storage Tanks
Reservoirs
Motors
Pumps
Valves
Material Type
Concrete
(m3)
3.2E-08
0
0
0
0
Steel
(kg)
6.4E-05
0
8.8E-06
0
0.0021
Earthworks
(m3)
0
2.9E-05
0
0
0
Electrical
steel (kg)
0
0
4.1E-05
0
0
Stainless
18/8 coil
(kg)
0
0
0
4.5E-06
0
Cast
Iron
(kg)
0
0
3.9E-05
6.0E-05
0
Aluminum
(kg)
0
0
2.4E-06
0
0
Copper
(kg)
0
0
7.1E-
06
0
0
Source: GCWWprimary data collection.
20
-------
3.5 LCA Limitations
While limitations of this study are discussed throughout this paper, some of the main limitations
that readers should understand when interpreting the data and findings are as follows:
• Water Quality Metrics. A unique aspect to this study is that detailed water metrics were
collected for source water quality, disinfection by-products, and pathogens. The majority
of metrics under these categories are not included in standard LCIA methods; therefore,
these metrics are excluded from the OpenLCA model and are not linked to changes in the
model. However, TTHM and cryptosporidium exposure are reported as results categories
under the distribution life cycle stage. They are reported under the distribution life cycle
stage as this is where human exposure to these pathogens and DBFs occurs.
• Infrastructure and Capital Equipment. While primary data were collected for
infrastructure at the drinking water plant, only material and installation burdens are
included. Assembly of the actual components (e.g., pumps, motors, tanks) and end-of-life
of the infrastructure are excluded due to lack of available data. Exclusion of assembly is
not the case for the piping (at plant and in distribution system), which does includes
impacts from assembly. Additionally, the infrastructure burdens are normalized over each
component's total lifetime assuming that the water delivered every year is 123,560,247
cubic meters, which was the volume of delivered to consumers in 2011. In actuality, there
would be differences in water delivered per year over time. The lifetimes assumed for
each component are estimates based on historical information of the GCWW facility;
however, the study does include a sensitivity analysis to look at a wider range of potential
lifetimes of infrastructure components.
• DWT Plant Electricity Consumption. Electricity consumption at the drinking water
treatment plant could not be split out by life cycle stages within the plant. Therefore, this
electricity consumption is reported as a separate life cycle stage in the results ("Pumping
Energy, at Drinking Water Treatment Plant"), when in reality it should be allocated
among DWT unit processes.
• Support Personnel Requirements. Support personnel requirements are included in the
cost analysis, but excluded from the LCA model. The energy and wastes associated with
research and development, sales, and administrative personnel or related activities are not
included.
• Transferability of Results. While this study is intended to inform decision-making for a
wide range of stakeholders, the data presented here relate to one representative facility.
Further work is recommended to understand the variability of key parameters across
specific situations.
• Representativeness of Background Data. Background processes are representative of
either U.S. average data (in the case of data from U.S. LCI) or European average (in the
case of ecoinvent) data. In some cases European ecoinvent processes were used to
represent U.S. inputs to the model (e.g., for chemical inputs) due to lack of available
representative U.S. processes for these inputs. The background data, however, met the
criteria listed in the project QAPP for completeness, representativeness, accuracy, and
reliability.
21
-------
• Representativeness of supply chains for sodium hexametaphosphate and sodium
fluorosilicate. LCI data for sodium hexametaphosphate or sodium fluorosilicate were not
available for use in this study. Therefore, surrogate processes of sodium tripolyphosphate
and hydrogen fluoride were used to model sodium hexametaphosphate and sodium
fluorosilicate respectively. These surrogates were chosen as the production processes for
these chemicals were closest to the actual chemicals used.
• Data Accuracy and Uncertainty. In a complex study with literally thousands of numeric
entries, the accuracy of the data and how it affects conclusions is truly a difficult subject,
and one that does not lend itself to standard error analysis techniques. The reader should
keep in mind the uncertainty associated with LCA models when interpreting the results.
Comparative conclusions should not be drawn based on small differences in impact
results.
4. BASE CASE COST ANALYSIS
The focus of the cost analysis is to understand the contribution of life cycle stages to the overall
cost of water delivered and, moving forward, to determine how different disinfection alternatives
impact the final cost of water to the consumer. The remainder of this section provides additional
details on the cost analysis data, methodology, and assumptions.13
4.1 Base Case Data Sources
The costs analysis used actual cost data provided by GCWW. GCWW provided annual operating
costs for the Richard Miller Treatment Plant from 2011 and capital improvement project costs
from 2000 to 2011. GCWW provided the treatment plant operating costs detailed by treatment
unit process.
4.1.1 Annual Operating Costs
Table 10 shows the costs included in each DWT stage. GCWW does not track maintenance for
the acquisition system separately. Therefore, costs were not allocated to the drinking water
acquisition process. In addition, many costs, such as operating and maintenance labor, are
incurred on a plant-wide basis. Therefore, a separate line item for overhead is included in the
costs. Table 10 also shows the total plant costs for both Base 1 and Base Case 2.
In addition to the cost data elements listed in Table 10, GCWW also provided information on
revenues and the price of drinking water to the consumer. EPA used these data elements to
evaluate how changes to the disinfection technology may change revenues and consumer prices.
These data were not used in the base case model.
13 All supporting data used in the cost analysis are included in a separate Excel file.
22
-------
Table 10. Annual Costs collected from GCWW.
Stage
Drinking Water
Acquisition
Drinking Water Treatment
Plant, Energy and
Infrastructure
Pre-Disinfection
Primary Disinfection
Distribution
Overhead
Unit Processes
Included in Base
Cases 1 and 2
Acquisition
Flocculation
Sedimentation
Filtration (sand)
GAC adsorption
Conditioning
Gaseous chlorine
Fluoridation
Distribution
Plant overhead for
all processes other
than primary
disinfection
Cost Elements
None
• Electricity for pumping3
• Chemicals (alum, polymer)
• Lime
• Sludge removal
• Sand replacement
• Sand disposal
• GAC replacement
• GAC regeneration (natural
gas, permit costs)b
• Chemicals (caustic soda,
sodium
hexametaphosphate)
• Chemicals (gaseous
chlorine)
• Maintenance0
• Chemicals
(hydrofluorosilicic acid)
• Chemicals (sodium
hypochlorite)
• Electricity"1
• Labor
• Maintenance
Total Base Case 1
Total Base Case 2 (no GAC)
Cost ($/year)
Included in overhead
$1,283,000
$983,000
$69,000
$30,000
$1,515,000
$898,000
$128,000
$365,000
$2,508,000
$2,212,000
$9,992,000
$8,477,000
a GCWW provided an annual amount of purchased electricity at the plant of 39,125,286 kWh and a unit cost of
electricity of $0.0328/kWh. EPA calculated the annual cost of electricity.
b GCWW estimated that 42,000 ccf of natural gas are required per regeneration of GAC and 18 regenerations occur
per year. GCWW provided a unit cost of natural gas of $0.0057/scf (1 ccf =100 scf). EPA calculated the annual cost
of natural gas.
0 Maintenance costs for the disinfection unit process are broken out separately from the overall maintenance costs to
evaluate potential changes for the alternative disinfection technology.
d GCWW provided an annual amount of electricity for distribution of 76,137,689 kWh and a unit cost of electricity
of $0.0328/kWh. EPA calculated the annual cost of electricity.
4.1.2 Capital Costs
GCWW provided data on capital improvement project (CIP) expenditures from 2000 to 2011 for
all of their operations (not limited to the Richard Miller Treatment Plant). GCWW provided the
capital spending data in two categories:
• Facilities, including water treatment plants, distribution pump stations, backup
generators, reservoirs, and storage tanks; and
• Water mains, including replacements and expansions.
23
-------
Table 11 summarizes the CIP spending from 2000 to 2011. As can be seen, yearly capital
spending can vary significantly depending on the extent and nature of capital improvement
projects. For example, the $14,855,000 CIP spending on facilities in 2011 covered 29 projects,
including: beginning construction of the UV treatment facility and replacing a portion of the
filter house roof at the Richard Miller Treatment Plant; repairing secondary clarifiers and
building a new sewer line at the Bolton Treatment Plant; and construction of a pump station,
backup generator, reservoir, and elevated storage tank along the distribution system.
From 2000 to 2011, the average annual facility CIP spending is $9,076,083 with a standard
deviation of $4,023,380, or about 44% of the average. This standard deviation reflects the large
variation from year to year.
GCWW has a goal of replacing 1% of water mains each year. In 2011, the 34.4 miles of water
main work encompassed 4.8 miles of new main extensions and 29.6 miles (or about 0.94%) of
water main replacement. The running average of water main replacement from 2000 to 2011 is
0.98% per year. From 2000 to 2011, the average annual water main CIP spending is $32,511,458
with an average of 44 miles of water main work per year. The standard deviation of spending is
$6,545,454, or about 20% of the average. The average spending per mile of water main work is
approximately $739,180 per mile.
Table 11. GCWW Capital Improvement Projects Spending for Facilities and Water Mains from
2000 to 2011
Year
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
Facilities Capital
Improvement Projects
CIP Spending
$ 14,855,000
$ 12,157,000
$ 5,889,000
$ 4,833,000
$ 3,985,000
$ 8,061,000
$ 7,936,000
$ 9,773,000
$ 13,197,000
$ 14,693,000
$ 9,856,000
$ 3,678,000
Water Main Installations Capital Improvement
Projects
Miles Completed
34.4
36.2
32.3
46.3
35.1
44.9
52.3
61.1
61.0
48.4
30.3
45.5
Water Main Design
Estimated Value
$ 33,207,825
$ 40,169,576
$ 40,997,569
$ 27,779,798
$ 35,469,183
$ 32,067,642
$ 22,707,669
$ 28,039,881
$ 35,999,391
$ 42,008,784
$ 26,693,046
$ 24,997,132
Source: GCWW Engineering Division Report to the Director.
4.2 Base Case Cost Method
EPA calculated the Base Case 1 and 2 costs directly from input provided by GCWW using the
steps below.
1. Match provided costs to the unit processes shown in Table 1. Note that GCWW did not
provide costs for ferric sulfate, which was only used 100 days in 2011. Costs for this
chemical are not included.
24
-------
2. Calculate the unit costs for items using the total annual costs and the quantities provided
(e.g., compute costs in $/pound chemical). Where GCWW provided average, minimum,
and maximum quantities, EPA calculated the unit cost using the average value. These
unit costs were not needed to compute total costs because GCWW provided the total
annual costs. However, EPA used unit costs in the sensitivity analyses and in the
evaluation of alternative disinfection technologies.
3. Calculate Base Case 1 and 2 totals using the costs provided. Base Case 2 does not include
GAC; therefore, EPA did not include the GAC replacement or regeneration costs in Base
Case 2.
4. Normalize the total costs to a cubic meter of drinking water delivered.
4.3 Cost Data Quality, Assumptions, and Limitations
As stated previously, all data used in the cost analysis were provided by GCWW and are for
calendar year 2011. The plant size and characteristics should be considered when translating
these costs to other DWT plants.
Because GCWW was not able to provide a breakout of labor by unit process, EPA used the total
labor costs for workers involved in all plant operations. These labor costs exclude personnel
involved in administration. However, administration costs and similar overhead that are not tied
directly to operations (e.g., administration personnel and expenses, office building utility bills,
insurance) are less likely to change in response to implementing new technologies in the DWT
plant. Therefore, all of the alternate technologies studied here are assumed to have similar
administration and overhead costs as the base case.
An important note is that plant labor is a significant component of the total plant costs and may
have the most variability between drinking water plants due to size and age differences.
5. BASE CASE RESULTS
Figure 3 displays the Base Case 1 contribution analysis results, Figure 4 displays more detailed
Base Case 1 results on the unit process level, Figure 5 presents comparative summary results by
life cycle stage for Base Case 1 versus Base Case 2, Figure 6 presents the percent change across
the impact results when adsorption is excluded, and Table 12 provides Base Case 1 and Base
Case 2 results per functional unit.14 This study was not able to collect data to determine whether
the use of GAC influences the effluent quality such as cryptosporidium and TTHM; therefore,
these categories are excluded from the Base Case 2 results' figures and tables.
Base case findings of note include:
14 The results for the life cycle assessment and cost analysis are presented in a separate Excel file.
25
-------
Base Case 1 shows slightly increased environmental impacts compared to Base Case 2
due to the addition of adsorption. Smog and human health criteria impacts are the most
sensitive to the difference.
Labor and energy costs are the largest contributions to DWT plant costs (18% and 38%,
respectively, for Base Case 1; and 21% and 45%, respectively, for Base Case 2; including
both plant and distribution costs).
Eliminating adsorption (including GAC production and regeneration) reduces total costs
by approximately 15 percent.
Disposal of sedimentation waste is the largest contributor to eutrophication potential
impacts (Figure 4). This is a result of the waterborne emissions of BOD, COD, and
ammonia leaching from the sedimentation waste (Table 4).
1.2 m3 of blue water are required to deliver 1 m3 of treated drinking water to the
consumer. A majority of the water loss occurs during the distribution stage. Based on
data collected from GCWW, 0.19 m3 of water is lost per m3 of drinking water delivered
during the distribution stage.
Global warming, energy demand, fossil depletion, acidification, human health cancer,
human health criteria and ecotoxicity impacts are largely driven by electricity
consumption at the drinking water treatment plant and during distribution to the
consumer.
Distribution is the largest contributor to metal depletion, accounting for 78 percent of
impacts. The distribution metal depletion is due primarily to the metal used in the iron
pipes throughout the distribution network infrastructure. Infrastructure at the DWT plant
and pre-disinfection account for 19 percent of metal depletion impacts, which is largely
attributable steel used for construction of the DWT plant and upstream infrastructure
required for production of the chemicals used during pre-disinfection processes (e.g.,
alum coagulant, sodium hexametaphosphate, sodium hydroxide, iron sulfate).
Overall, the primary disinfection with gaseous chlorine life cycle stage only contributes
zero to five percent to the total life cycle impacts of DWT for the results categories
examined.
26
-------
0%
Cost
Cryptosporidium
TTHM
Chlorine Usage
Global Wanning
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion |
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
Percent Contribution by Life Cycle Stage
20% 40% 60% 80%
100%
• Source Water Acquisition
• Drinking Water Treatment Plant Energy Usage
n Pre-Disinfection
• Primary Disinfection
D Distribution
Overhead*
* Overhead is only considered as stage for cost results category.
Figure 3. Base Case 1 contribution analysis results.
27
-------
Percent Contribution by Unit Process
20% 40% 60%
80%
Cost
Cryptosporidium
TTHM
Chlorine Usage
Global Wanning
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
100%
• Source Water Acquisition
• Flocculation
• Filtration
• Conditioning
Fluoridation
Overhead*
l Drinking Water Treatment Plant Energy Usage
I Sedimentation
Adsorption
Primary Disinfection
Distribution (Excluding Fluoridation)
* Overhead is only considered as stage for cost results category.
Figure 4. Base Case 1 contribution analysis results with unit process detail.
28
-------
I Source Water Acquisition
Distribution
Pre-Disinfection
l Drinking Water Treatment Plant Energy Usage
Overhead
I Primary Disinfection
20%
Percent of Maximum
40% 60%
80%
100%
Figure 5. Base Case 1 and Base Case 2 comparative summary results.
29
-------
-16% -14% -12%
Percent Change
-10% -8% -6%
-4%
Cost
Smog
Human Health, Criteria
Fossil Depletion
Energy Demand
Acidification
Eutrophication
Global Warming
Metal Depletion
Human Health, Cancer
Ozone Depletion
Human Health, NonCancer
Blue Water Use
Chlorine Usage
Ecotoxicity
Figure 6. Percent change in impacts if adsorption is excluded.
Table 12. Base Case 1 and Base Case 2 results per m3 drinking water delivered to the consumer.
Results Category
Cost
Cryptosporidium
TTHM
Chlorine Usage
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer, Total
Human Health, NonCancer, Total
Human Health, Criteria
Ecotoxicity, Total
Unit
$
oocyst
kg TTHM
kgCl2
kg C02 eq
MJ
kg oil eq
kg H+ mole eq
kgN eq
m3
kg O3 eq
kgCFC-lleq
kg Fe eq
CTU
CTU
kgPMWeq
CTU
Base Case 1
0.081
1.00
1.6E-05
0.0018
1.04
19.8
0.36
0.48
9.7E-04
1.20
0.067
2.8E-08
0.036
2.9E-11
3.2E-11
0.0015
4.4E-04
Base Case 2
0.069
0.0018
1.03
19.5
0.35
0.47
9.6E-04
1.20
0.063
2.8E-08
0.035
2.9E-11
3.2E-11
0.0014
4.4E-04
30
-------
5.1 Base Case Normalized Results
Figure 7 displays the Base Case 1 normalized results. Larger sections of the chart indicate those
impacts where DWT makes relatively larger contributions to national per capita impacts. Impacts
related to fossil fuel combustion from electricity such as acidification potential, smog formation
potential, global warming potential, and human health criteria are relatively high. Eutrophication
impacts are also relatively high, primarily due to the disposal of the sedimentation waste. Other
metrics such as ozone depletion potential, ecotoxicity, human health cancer and noncancer are
relatively low.
HHNC: 0.01%,
HHCa: 0.4
OOP: 0.1%^^ HHCr
ETP: 0.002%
I Global Warming (GWP)
I Acidification (AP)
lEutrophication (EP)
I Smog (SFP)
1 Ozone Depletion (OOP)
] Human Health, Cancer (HHCa)
Human Health, NonCancer (HHNC)
Human Health, Criteria (HHCr)
Ecotoxicity (ETP)
Figure 7. Base case normalized results.
5.2 Infrastructure Contribution to Base Case Results
Table 13 and Figure 8 display the contribution of infrastructure at the drinking water treatment
plant and in the distribution system to the Base Case 1 results. For the majority of impact
categories, the distribution pipe network is the infrastructure component with the highest
impacts. For the majority of impact categories, excluding metal depletion and ecotoxicity,
infrastructure contributes 5% or less to the total impacts. Metal depletion, however, is largely
driven by infrastructure, with infrastructure from the drinking water treatment plant and
distribution system accounting for approximately 68% of all metal depletion impacts. The
remaining metal depletion impacts are also primarily due to upstream infrastructure impacts, for
instance from the construction of plants which produce chemicals used for water treatment.
-------
i Operational Impacts
I Infrastructure Impacts
0%
20%
40%
60%
80%
100%
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
Figure 8. Infrastructure contribution analysis.
32
-------
Table 13. Contribution of infrastructure to Base Case 1 results per m3 drinking water delivered to the consumer.
Life Cycle Stage
Source Water
Acquisition
Pre-Disinfection
Primary
Disinfection
Distribution
Total
Subprocess
Source water acquisition
infrastructure
Conditioning
infrastructure
Adsorption
infrastructure
Filtration infrastructure
Lime addition
infrastructure
Sedimentation
infrastructure
Flocculation
infrastructure
Primary disinfection
infrastructure
Pipe Network
Water Storage
Valves
Pumps
Motors
All
% of total impact
Impact Category
Global
Warming
kg CO 2
eq
8.4E-04
3.6E-04
0.0012
4.5E-04
2.3E-04
2.0E-04
6.9E-04
2.3E-05
0.011
1.6E-04
0.0043
1.3E-04
1.5E-04
0.020
1.9%
Energy
Demand
MJ
0.010
0.0033
0.011
0.0041
0.0019
0.0033
0.0069
2.1E-04
0.26
0.0025
0.069
0.0022
0.0026
0.37
1.9%
Fossil
Depletion
kg oil eq
2.0E-04
6.4E-05
2.2E-04
8.1E-05
3.7E-05
7.1E-05
1.4E-04
4.1E-06
0.0057
5.0E-05
0.0014
4.4E-05
4.8E-05
0.0081
2.3%
Acidification
kg H+ mole
eq
1.4E-04
5.2E-05
1.7E-04
6.5E-05
3.2E-05
6.2E-05
1.1E-04
3.3E-06
0.0036
3.9E-05
9.1E-04
2.9E-05
8.3E-05
0.0053
1.1%
Eutrophication
kgN eq
1.5E-07
5.2E-08
1.7E-07
6.5E-08
3.4E-08
8.3E-08
l.OE-07
3.3E-09
2.9E-06
4.1E-08
7.4E-07
2.3E-08
4.6E-08
4.4E-06
0.5%
Blue
Water
Use
m3
4.8E-06
4.6E-06
1.6E-05
5.8E-06
2.9E-06
4.3E-07
8.4E-06
2.9E-07
3.2E-05
1.6E-06
4.6E-05
9.7E-07
1.4E-06
1.2E-04
0.0%
Smog
kg O3 eq
5.1E-05
1.9E-05
6.3E-05
2.4E-05
1.3E-05
2.8E-05
3.9E-05
1.2E-06
0.0015
1.3E-05
2.3E-04
7.4E-06
1.2E-05
0.0020
3.0%
Ozone
Depletion
kgCFCll
eq
6.5E-11
1.6E-11
5.5E-11
2.0E-11
l.OE-11
2.0E-11
3.0E-11
l.OE-12
1.7E-10
l.OE-11
2.2E-10
6.1E-12
9.6E-12
6.3E-10
2.3%
Metal
Depletion
kg Fe eq
7.2E-05
7.3E-05
2.5E-04
9.1E-05
1.9E-05
5.5E-06
1.3E-04
4.7E-06
0.015
2.3E-04
0.0076
1.4E-04
3.5E-04
0.024
67.3%
Human
Health,
Cancer,
Total
CTU
7.6E-14
2.4E-14
8.1E-14
3.0E-14
l.OE-14
1.3E-14
4.4E-14
1.5E-15
1.1E-13
1.8E-14
5.6E-13
2.4E-14
3.0E-14
l.OE-12
3.5%
Human
Health,
NonCancer,
Total
CTU
3.0E-14
1.4E-14
4.8E-14
1.8E-14
9.0E-15
4.4E-15
2.6E-14
9.0E-16
3.3E-14
3.3E-14
1.1E-12
4.9E-15
1.6E-12
2.9E-12
9.0%
Human
Health,
Criteria
kgPMW
eq
7.0E-07
4.4E-07
1.5E-06
5.5E-07
1.7E-07
2.4E-07
8.2E-07
2.8E-08
7.8E-06
5.4E-07
1.6E-05
5.0E-07
7.7E-07
3.0E-05
2.0%
Ecotoxicity,
total
CTU
1.7E-06
3.3E-07
1.1E-06
4.1E-07
1.6E-07
4.3E-07
6.6E-07
2.1E-08
1.7E-05
3.0E-07
8.8E-06
3.9E-07
2.5E-06
3.4E-05
7.7%
33
-------
6.
BASELINE SENSITIVITY ANALYSES
To see the influence of the assumptions made in an LCA model, it is important to conduct
sensitivity analyses. To carry out such an analysis, the assumption of interest is changed and the
entire LCA is recalculated. In this study, EPA conducted sensitivity analyses for key base case
assumptions. Table 14 shows the sensitivity analyses for the base case, the values used, and
whether LCA or cost results were generated for the sensitivity. Costs results were generated if
changes to the LCA parameter could impact the costs. For example, changing the quantity of
chlorine used at the plant would change the costs. On the other hand, varying the quantity of
cryptosporidium in the final drinking water would not result in cost changes if no changes to the
plant were made. Table 15 provides the electricity grid fuel mix used in both the baseline and the
sensitivity analysis.
Table 14. Sensitivity analyses for base case model runs.
Parameter
Chlorine usage
Lime consumption
Alum coagulant usage
Sodium hypochlorite usage
during distribution
Natural gas for GAC
reactivation
DBF exposure
Cryptosporidium exposure
Electricity usage at plant a
Electricity usage during
distribution a
Electricity unit cost (plant
and distribution)
Electricity grid
Lifetime of DWTP
infrastructure components
Lifetime of DWT
distribution system
infrastructure components
Values
Minimum, maximum, and average
values obtained from GCWW
Minimum, maximum, and average
values obtained from GCWW
Minimum, maximum, and average
values obtained from GCWW
Minimum, maximum, and average
values obtained from GCWW
Minimum, maximum, and average
values obtained from GCWW
Minimum, maximum, and average
values obtained from GCWW
Minimum, maximum, and average
values obtained from GCWW
±10% of value obtained from
GCWW
±10% of value obtained from
GCWW
±20% of value obtained from
GCWW
Average U.S. grid, RFCW NERC
regional grid
±25 years for buildings, pipes, and
other features (baseline =100 years)
±25 years for buildings, pipes, and
other features (baseline =100
years); ±10 years for pumps and
motors (baseline = 25 years)
LCA
Results
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Cost
Results
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Yes
No
No
No
a Varying the electricity usage by ± 10% also provides an indication of the effects of varying
the total cost of electricity by ±10%. EPA also varied total electricity costs by ±20% (plant
and distribution) as shown on the cost results worksheet.
34
-------
Table 15. U.S. electrical grid fuel i
Electricity source
Bituminous coal
Lignite coal
Natural gas
Distillate oil
Residual oil
Biomass
Nuclear
Hydro
Wind
Solar
Geothermal
MSW, non-biogenic
Petroleum coke
Petroleum waste oil
Tire derived fuel
Other fuels
Other gases
U.S. Average
46.24%
1.96%
21.43%
0.18%
0.57%
1.33%
19.57%
6.03%
1.34%
0.021%
0.36%
0.15%
0.35%
0.022%
0.030%
0.072%
0.28%
profiles15
RFCW NERC
Region
77.06%
0%
2.41%
0.14%
0.0024%
0.48%
18.24%
0.62%
0.20%
0%
0%
0.028%
0.20%
0.0014%
0.0083%
0.060%
0.54%
Sensitivity analyses findings of note include:
• As displayed in Figure 9, the use of the U.S. average grid electricity mix resulted in
considerably lower global warming, smog, and acidification impacts compared to use of
the ReliabilityFirst Corporation West (RFCW) grid, which is the North American
Electrical Reliability Corporation (NERC) region the GCWW Richard Miller Treatment
Plant is located. This is largely due to the higher use of coal in the RFCW grid compared
to the U.S. average grid. However, use of the RFCW grid electricity mix significantly
reduced Human health cancer and ecotoxicity impacts, which is due to the lower natural
gas usage in the RFCW grid mix compared to the U.S. average grid mix.
• Figure 10 shows the results of eight cost sensitivity analyses in terms of percent change
from the baseline. Labor and energy are the highest contributors to the overall plant costs,
so changes in chemical quantities and costs generally do not have a significant impact on
the overall costs. Cost results are, however, sensitive to the electricity unit cost.
• Increases/decreases in plant electricity usage had the most effect on impacts associated
with fossil fuel production and combustion such as global warming potential, human
health criteria, smog, acidification, fossil depletion, human health cancer and energy
demand (Figure 11).
• Impact results vary +/- zero to six percent when varying the distribution electricity usage
+/- 10 percent (Figure 12). Impacts related to fossil fuel combustion (e.g., global
warming, energy demand, fossil depletion, acidification) are most affected. These results
clearly show the DWT model is sensitive to the electricity usage during distribution, and
eGRID 2008 (Emissions and Generation Resource Integrated Database). U.S. EPA.
(www.epa.gov/cleanenergy/egrid).
35
-------
that electricity usage during distribution is an impactful process in the overall DWT life
cycle. The model is more sensitive to varying electricity usage during distribution
compared to varying electricity usage at the DWTP since distribution requires almost
twice as much electricity compared to treatment at the plant.
Crypto sensitivity results show that exposure results vary on average approximately -15
percent/+10% from Base Case 1.
TTHM sensitivity results indicate that exposure results vary on average approximately -
50 percent/+32 percent from Base Case 1.
Chlorine usage results vary on average approximately -27 percent/+46 percent from Base
Case 1 (Figure 13). No other impact categories are sensitive to the chlorine usage range.
Results of the infrastructure sensitivity analysis are displayed in Figure 14 (infrastructure
at DWTP) and Figure 15 (infrastructure for distribution system). The lifetimes assumed
for each infrastructure component are estimates based on historical information of the
GCWW facility (100 years for buildings, pipes, other features, and 25 years for pumps
and motors); however, the study does include a sensitivity analysis to look at a wider
range of potential lifetimes of infrastructure components. For building, pipes and other
features (e.g., tanks and reservoirs) the lifetime is varied +/- 25 years, while for the
pumps and motors, the lifetime is varied +/- 10 years. Overall life cycle impacts increase
with a decrease in the infrastructure lifetime, since the infrastructure burdens are
normalized over less total water delivered. The infrastructure lifetime is only sensitive to
the metal depletion category, since this is the only impact category in which
infrastructure is a significant component. All other impact categories vary less than 5
percent from the base case for this sensitivity analysis. The distribution system
infrastructure has the greatest impact on metal depletion results as approximately 95
percent of the 3,135 miles of distribution system piping is iron.
Figure 16, Figure 17, Figure 18, and Figure 19 show the sensitivity analysis results for
alum coagulant usage during flocculation, lime usage during sedimentation, sodium
hypochlorite usage during distribution, and natural gas consumption for GAC
reactivation respectively. The LCA model is not sensitive to these parameters within the
potential operational range supplied by GCWW. Cost results vary the most for the input
quantity of alum coagulant.
36
-------
Percent Change
-60% -50% -40% -30% -20% -10% 0%
10% 20%
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
• Acquisition & treatment
d Distribution
L
40%
Figure 9. Significance of electricity mix: RFC West versus U.S. average baseline.
Relative Change in Total Costs for Base Case (%)
-8.00 -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00
Electricity Unit Cost ($/kWh)
Electricity (distribution) (purchased) (kWh)
Alum Usage (lb/1,000 ft3)
Natural Gas for GAC Regeneration Unit Cost ($/scf)
Electricity (at plant) (purchased) (kWh)
Lime Usage (lb/1,000 ft3)
Gaseous Chlorine Usage (lb/1,000 ft3)
Sodium Hypochlorite Usage (lb/1,000 ft3)
Figure 10. Tornado chart of the sensitivity analysis results for the relative changes in total costs for
Base Case 1.
37
-------
-4%
Percent Change
-3% -2% -1% 0% 1% 2% 3% 4%
Cost
Cryptosporidium
TTHM
Chlorine Usage
Global Wanning
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
I Minimum DWTP Electricity Usage (-10%)
I Maximum DWTP Electricity Usage (+10%)
Figure 11. Base Case 1 DWTP electricity usage sensitivity analysis.
Percent Change
-8% -6% -4% -2% 0% 2% 4%
Cost
Cryptosporidium
TTHM
Chlorine Usage
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
6% 8% 10%
D Minimum Distribution
Electricity Usage (-10%)
• Maximum Distribution
Electricity Usage (+10%)
Figure 12. Base case 1 distribution system electricity usage sensitivity analysis.
38
-------
Percent Change
-40% -30% -20% -10% 0%
Cost
Chlorine Usage
Global Wanning
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
t
10% 20% 30% 40% 50% 60%
—I 1 1
I Minimum Chlorine Usage
I Maximum Chlorine Usage
Figure 13. Base Case 1 chlorine usage sensitivity analysis.
Percent Change
-0.6% -0.4% -0.2% 0.0% 0.2% 0.4%
0.6%
0.8%
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
T
• Minimum DWTP Lifetime
• Maximum DWTP Lifetime
Figure 14. Base Case 1 DWTP infrastructure lifetime sensitivity analysis.
39
-------
Percent Change
-15% -10%
-5%
0%
5%
10%
15%
20%
25%
1 Minimum Distribution System
Lifetime
• Maximum Distribution System
Lifetime
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
Figure 15. Base Case 1 distribution system infrastructure lifetime sensitivity analysis.
-3%
Cost
Chlorine Usage
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
-2%
Percent Change
-1% 0% 1%
L
i
j
2%
3%
4%
I Minimum Alum
Usage
I Maximum Alum
Usage
Figure 16. Base Case 1 alum coagulant usage sensitivity analysis.
40
-------
-0.5%
Percent Change
0.0%
0.5%
Cost
Chlorine Usage
Global Wanning
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
1 Minimum Lime Usage
1 Maximum Lime
Usage
J
Figure 17. Base Case 1 lime usage sensitivity analysis.
-0.5%
Percent Change
0.0%
0.5%
1.0%
I Minimum Sodium
Hypochlorite Usage
I Maximum Sodium
Hypochlorite Usage
Cost
Chlorine Usage
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
Figure 18. Base Case 1 distribution sodium hypochlorite usage sensitivity analysis.
41
-------
-1.0%
-0.5%
Percent Change
0.0%
0.5%
1.0%
Cost
Cryptosporidium
TTHM
Chlorine Usage
Global Wanning
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
T
I Minimum Natural Gas
Usage
I Maximum Natural Gas
Usage
7.
Figure 19. Base Case 1 natural gas for GAC reactivation sensitivity analysis.
IN-PLANT ALTERNATIVE DISINFECTION TECHNOLOGIES
EPA investigated the use of the following in-plant disinfection alternatives by collecting data
from the EPA partners noted.
• Conventional mercury-vapor ultraviolet (UV) disinfection (Aquionics)
• LED UV disinfection (Aquionics)
• Ferrate disinfection (Ferrate Treatment Technologies, LLC (FTT))
EPA had intended to include plasma-bead UV disinfection from Imaging Systems Technology
(1ST) in this study. While EPA investigated plasma-bead UV technologies and collected
information from 1ST, it was determined that this technology is still too early in development to
model quantitatively.
Table 16 shows where each technology is implemented in the drinking water model plant. The
following subsections describe each technology, EPA's data collection for the technology, the
methodology used to compare impacts and costs for the alternative technology, and the final
results.
42
-------
Table 16. Unit process matrix for alternative disinfection technologies.
Life Cycle Stage Reported
Source Water Acquisition
Drinking Water Treatment
Plant, Energy and Infrastructure
Pre-Disinfection
Primary Disinfection
Distribution
Use
Unit Processes Covered
Source Water Acquisition
Drinking Water Treatment Plant, Energy and
Infrastructure
Flocculation
Alum Coagulant
Sedimentation
Ferrate oxidation
Disposal, Sedimentation Waste
Filtration
Adsorption
GAC Production
GAC Regeneration
Primary Disinfection, Gaseous Chlorine
Alternate UV Disinfection (conventional,
LED)
Ferrate disinfection
Fluoridation
Transport, Treated Drinking Water, Water
Supply Pipeline
Distribution Infrastructure, Drinking Water
Drinking Water Consumption
-------
Sedimentation
Disposal of
Waste
LE
GEND
C )
L>
Primary Input/
Final Demand
Primary Process
Reference
Supply Chain
Or gate (multiple
outputs)
Or gate (multiple
inputs)
>
Primary
Disinfection,
Gaseous Chlorine
Conditioning U-
£
*
Primary
Disinfection,
Conventional UV
-x
~~V
Primary
Disinfection, LED
UV
i
Primary
Disinfection,
Ferrate
i
Fluoridation
Figure 20. System boundaries of drinking water treatment base case and in-plant disinfection
alternatives.
7.2 Aquionics Conventional UV Disinfection System
Aquionics provides UV disinfection equipment for drinking water and municipal treatment
plants. EPA has partnered with Aquionics to provide data on conventional and LED UV for this
study. Conventional UV uses mercury-vapor lamps to provide the UV light which de-activates
microorganisms. UV can also breakdown unwanted chemicals such as organic compounds.
According to Aquionics product information, Aquionics UV systems use low wavelength UV
44
-------
light, which can breakdown total organic carbon (TOC) molecules into smaller compounds that
can then be removed by other unit processes.16 Potential benefits of UV include:
• Removes chlorine-resistant pathogens;
• Reduces chlorine quantities required on site; and
• Reduces generation of DBFs.
EPA collected data on conventional UV directly from Aquionics.
7.2.1 LCA Model
For the conventional UV LCA model, only the primary disinfection stage is changed from the
base case DWT model. Aquionics provided electricity and infrastructure data to assist in
developing the LCA model. EPA made the following assumptions for the conventional UV
model based on the data provided by Aquionics:
• For primary disinfection, a 3.6 MOD UV unit was modeled. Therefore, 30 active
units and one backup unit would be required for disinfection at the 106 MOD
GCWW base case facility.
• Conventional mercury-vapor UV disinfection requires 0.042 kWh electricity per
cubic meter of water treated. This value was calculated using Aquionics' estimate
that 210,240 kWh/yr of electricity are required to treat 3.6 MOD of water at a
dose of 40 ml/cm2.
• Each 3.6 million gallon per day (568 m3 per hour) UV disinfection unit consists of
six mercury-vapor bulbs. The lamp lifetime is approximately 8,000 hours.
• The weight of each disinfection unit is 288 Ib. The unit consists of a stainless steel
vessel, quartz sleeves for the lamps, electronics for control units, synthetic rubber
for wiper rings, and the mercury vapor lamps. Aquionics estimated a lifetime for
each part (see Table 18). With the exception of the lamps, EPA used the average
lifetime of the disinfection components of five years for the LCA model for
simplicity. The conventional UV unit infrastructure has a negligible impact on the
LCA results; therefore, the results are not sensitive to this assumption.
• Disinfection with conventional UV does not result in any formation of DBFs.
• Disinfection with conventional UV can lead to the same levels of cryptosporidium
reduction as disinfection with gaseous chlorine.
EPA made some additional assumptions to complete the conventional UV LCA model:
• No chlorine is required for primary disinfection, but some gaseous chlorine is still
required to maintain a chlorine residual in the distribution system. This is based
on information from FTT (see Section 7.4). The sodium hypochlorite added
during distribution is still added in the same amount as this is required to boost
chlorine residual in certain parts of the distribution system. This sodium
hypochlorite boost may not be applicable for other drinking water systems.
16 Aquionics website (www.aquionics.com)
45
-------
• EPA assumed there was 400 mg mercury per lamp.17
• Other components of the UV lamp were modeled based on an amalgam lamp: 7.5
mm3 argon/lamp, 200 mg indium/lamp, 200 mg molybdenum/lamp, 4 g soldering
materials/lamp, 20 g ceramics/lamp, and 300 g glass (assumed to be quartz
sleeve)/lamp. 8 The total weight of one lamp was therefore assumed to be
321.1305 grams.
• Besides the weight of the lamp, no data were available on the weight breakdown
of components of the 288 Ib UV unit; therefore, the following assumptions were
made: the stainless steel vessel accounts for 85% of the unit weight, the
electronics account for 13% of the weight, the synthetic rubber wiper rings
account for 0.5% of the total unit weight, and the remainder of the unit weight
(1.47%) is from the lamps.
Because the conventional UV lamps include mercury, which is considered a hazardous material,
the "chlorine usage" category from the base case analysis is expanded here to "hazardous
materials" to account for this mercury. This unique flow, which is not a typical category in LCA
studies, is only tracked based on data reported by data providers for specific life cycle stages, and
does not cover all potential upstream hazardous materials. However, this category aids in
understanding the hazardous materials at the drinking water treatment plant that workers may be
exposed to.
7.2.1.1 Unit Processes
The specific unit processes added for the conventional UV LCA model are identified below.
Disinfection
1. Primary Disinfection, Conventional UV. Primary disinfection with conventional
(mercury-vapor) UV. The inputs to this unit process include operation and infrastructure
requirements for the UV units.
2. Conventional UV Drinking Water Treatment, Operation. This process covers
electricity usage associated with operation of the UV units.
3. Conventional UV Drinking Water Treatment, Infrastructure. Infrastructure inputs
for the UV units are aggregated in this unit process.
4. Conventional UV Lamp. Represents infrastructure requirements for the mercury-UV
lamp and quartz sleeve encompassing the lamp.
5. Stainless Steel UV Vessel. Production of the stainless steel UV vessel.
6. Wiper Ring. Covers infrastructure requirements for synthetic rubber wiper rings.
7. Conventional UV Electronic Control Unit. Production requirements for an electronics
control unit.
17 Malley, J.P., Jr. 2006. Development of Standard Operating Plans for Mercury Release from UV Technologies.
Used in Drinking Water Treatment Plants. Course Lecture Materials University of New Hampshire, Durham, NH
18 Ekwall, Cecilia. 2004. LCA of tap water disinfection - a comparison of chlorine and UV-light. Department of
Biometry and Engineering, Swedish University of Agricultural Sciences. http://ex-
epsilon.slu.se/archive/00000280/01/cecilia ekwall 0402.pdf
46
-------
Use
8. Drinking Water Consumption, Conventional UV. Final delivery of water, which is
disinfected with conventional UV, to an average consumer. This unit process aggregates
the other main life cycle stages and is used to build the final product system. There are no
actual impacts associated with the drinking water consumption life cycle stage itself.
Table 17 displays the data sources used for the conventional UV model in addition to the data
sources used in the base case model (See Table 3). In general, data from Aquionics were used
where available. For upstream processes that would not be known by Aquionics, such as
information on production of UV lamp materials (e.g., mercury, molybdenum, glass), EPA used
information from ecoinvent v2.2. Data sets from ecoinvent v2.2 have not been adapted for this
project.
Table 17. Conventional UV data sources.
Process Data Source
Conventional UV disinfection operation Data Collection-Aquionics
Infrastructure for UV unit Data Collection-Aquionics, and
1718
assumptions from secondary sources '
Stainless steel for UV vessel ecoinvent v2.2
Mercury for UV lamp ecoinvent v2.2
Molybdenum for UV lamp ecoinvent v2.2
Ceramics for UV lamp ecoinvent v2.2
Glass (i.e., quartz sleeve) for UV lamp ecoinvent v2.2
Argon for UV lamp ecoinvent v2.2
Indium for UV lamp ecoinvent v2.2
Electronics module ecoinvent v2.2
Synthetic rubber for wiper rings ecoinvent v2.2
7.2.2 Cost A nalysis
Table 18 lists information Aquionics provided for use in the cost analysis.19 After adding in the
conventional UV system (including electricity usage, parts replacement, and amortized capital
investment) and reducing the gaseous chlorine usage, the total annual cost is $10,666,000, an
increase of $674,000 from Base Case 1.
19 Input data, calculations, and results for the UV cost analysis are included in the supporting Excel file.
47
-------
Table 18. Cost data provided by Aquionics.
Cost Element
UV unit6
PLC and Control Cabinet
Online UVT Monitor
Lamps0
Quartz Sleeves
Wiper Rings
UV Sensors
Lifetime (years)
20
20
20
0.9
5
2
8
Cost
$45,050/unit
$90,100/unit
$7,500/unit
$450/1 amp
$70/sleeve
$18/ring
$500/sensor
a See Excel worksheets for detailed cost calculations.
b Costs are for one unit, which can treat 3.6 MOD. EPA scaled the 3.6 MOD system to the
base case plant size and included one back-up unit.
0 8,000 hours equals approximately 0.9 years.
In addition, EPA included the following information and assumptions based on information
provided by Aquionics:
1. The Aquionics system is for a 3.6 MGD small-scale system. As described Section 7.2.1,
the Aquionics system was scaled up to 30 active units plus one backup unit for
disinfection at the 106 MGD GCWW base case facility. This scale-up factor of 31 was
applied to the capital costs of the UV unit and the PLC and control cabinet. Only one
UVT monitor is required. The resulting total capital equipment cost is $4,200,000.
2. Costs include replacement parts (lamps, quartz sleeves, wiper rings, and UV sensors).
Aquionics provided the cost and lifetime of each part and noted each UV unit uses six
lamps. EPA assumed each lamp requires one quartz sleeve, one wiper ring, and one UV
sensor.
3. Cost multipliers are often applied to equipment costs to account for other direct costs
such as installation, site work, and ancillary equipment and indirect costs such as
permitting, monitoring, and training. This study assumes Aquionics would provide piping
and electrical equipment required for the UV system. A contingency of 25% of total
capital equipment costs was included. The resulting total capital investment is
$5,250,000.
4. EPA amortized the total capital costs over the 20-year expected lifetime of a UV system
using a bond rate of 6 percent. The resulting annual, amortized cost is $457,000.
5. EPA did not include cost credits for any equipment that is no longer required with use of
the UV system. Plants may be able to reduce equipment required for chlorine addition
because less chlorine is required.
7.2.3 Results
Table 19 displays results of the conventional UV analysis on the basis of 1 m3 water delivered to
the consumer. Figure 21 presents comparative summary results by life cycle stage for Base Case
48
-------
1 versus conventional UV disinfection. Figure 22 presents the percent change across the impact
results when using conventional UV disinfection rather than gaseous chlorine disinfection.
Section 7.6 presents results comparisons between the three alternative disinfection technologies
and the base case. Some findings to note for the conventional UV results as compared to the base
case:
• Application of conventional UV technology increases impacts during disinfection
through increased electricity consumption and through new capital investment,
but eliminates the formation of disinfection by-products and greatly reduces
hazardous chlorine usage.
• With the exception of hazardous materials and DBF formation (i.e., TTHM results
category), the choice of disinfection technology does not significantly impact
overall life cycle results, since most impacts are driven by energy consumption
for pumping at the DWT plant and during distribution. This pumping energy
consumption is not affected by choice of disinfection technology.
• For the hazardous materials category, the quantity of mercury from the bulbs is
negligible compared to the quantity of chlorine used to maintain a residual in the
distribution network. This study does not distinguish between different hazard
levels of chlorine versus mercury in the "hazardous materials" results category.
Table 19. Conventional UV results per m3 drinking water delivered to the consumer.
Results Category
Cost
Cryptosporidium
TTHM
Hazardous Materials
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer, Total
Human Health, NonCancer, Total
Human Health, Criteria
Ecotoxicity, Total
Unit
$
oocyst
kg TTHM
kg C12 & HZ
kg CO2 eq
MJ
kg oil eq
kg H+ mole eq
kgN eq
m3
kg O3 eq
kgCFC-lleq
kg Fe eq
CTU
CTU
kgPMWeq
CTU
Conventional UV
$0.086
1.00
0
4.4E-04
1.07
20.4
0.37
0.49
9. 7E-04
1.20
0.069
2.8E-08
0.036
3.0E-11
3. IE- 11
0.0015
4.5E-04
49
-------
I Source Water Acquisition
Distribution
Pre-Disinfection
I Drinking Water Treatment Plant Energy Usage
Overhead
l Primary Disinfection
20%
Percent of Maximum
40%
60%
80%
100%
Figure 21. Base Case 1 and conventional UV comparative summary results.
50
-------
-80% -70% -60% -50%
Percent Change
-40% -30% -20%
-10% 0% 10% 20%
Chlorine Usage
Human Health, NonCancer
Metal Depletion
Blue Water Use
Eutrophication
Ecotoxicity
Human Health, Cancer
Ozone Depletion
Human Health, Criteria
Smog
Acidification
Energy Demand
Global Warming
Fossil Depletion
Cost
Figure 22. Percent change in impacts if using conventional UV rather than gaseous chlorine for
disinfection.
For most impacts examined, the increase seen for utilization of conventional UV versus gaseous
chlorine for primary disinfection is due to increased electricity usage. Therefore, a sensitivity
analysis is run here varying the electricity usage for conventional UV disinfection +/- 25 percent.
Figure 23 presents the results of this analysis. Overall, the total life cycle impacts for DWT
disinfection with conventional UV do not vary more than +/- 0.9% when varying the electricity
usage for conventional UV operation +/- 25 percent. This is primarily a result of the overall
small impact of the primary disinfection life cycle stage as compared to other life cycle stages
that are larger consumers of energy (e.g., pumping at the DWT plant, distribution of the treated
water to the consumer).
Figure 24 presents a tornado chart that displays the results of the total cost sensitivity analysis.
The cost sensitivity analysis performed a Monte Carlo simulation, varying the following:
• Amount of electricity required by the conventional UV system by ±25% (same as
was performed for the LCA sensitivity analysis).
• Cost of a conventional UV unit by ±10% ($45,050/unit).
• Bond rate from four to eight percent (a ±33% change from the baseline value of
six percent).
51
-------
The sensitivity results show that the total costs are most sensitive to the bond rate and closely
followed by the amount of electricity required by the conventional UV system. However, the
total cost changes are within approximately ±0.6%. Therefore, although the total costs are more
sensitive to the bond rate and electricity required by the UV unit than they are to the capital
equipment cost, the total costs are negligibly changed by the parameter values studied. This
result is expected as the disinfection costs are less than 8% of the total costs. Therefore, changes
to disinfection costs have a smaller impact on the total costs compared to the larger costs: pre-
disinfection (33%); distribution (27%); overhead (21%); and plant energy (12%).
• Primary Disinfection
Percent Change
-1.0% -0.8% -0.6% -0.4% -0.2% 0.0% 0.2%
0.4% 0.6% 0.8% 1.0%
Figure 23. Conventional UV electricity usage sensitivity analysis.
52
-------
Relative Change in Total Costs for Conventional UV Scenario (%)
-0.80 -0.60 -0.40 -0.20 0.00 0.20 0.40 0.60 0.80
Bond Rate (%)
UV Electricity Usage to treat 3.6 MOD (kWh/yr)
UV System Cost per Units needed to treat 3.6 MOD
($)
Figure 24. Tornado chart of the sensitivity results for the relative changes in total costs for the
conventional UV scenario.
7.3 Aquionics LED UV Disinfection System
Aquionics offers a line of UV-LED disinfection equipment, which provides the same benefits of
conventional UV but uses an LED light source rather than a mercury lamp. The LED UV system
Aquionics has developed is for point-of-use applications (e.g., for laboratory equipment, health
care equipment, stand-alone point-of-use). EPA made some assumptions to scale this technology
to the 106 MGD system from the base case, but it is important to note that such large-scale LED
UV disinfection technology does not currently exist. Utilization of LED for small-scale point-of-
use applications is examined in Section 8. Cost data were not available for LED UV disinfection
and are not included in this analysis.
7.3.1 LCA Model
EPA made the following assumptions for the in-plant LED UV analysis:
• For the LED UV LCA model, only the primary disinfection stage is changed from
the base case DWT model.
• Water treated per disinfection unit is assumed equivalent to that treated under the
conventional UV scenario (since no large-scale LED UV system exists, primary
data on water treated for large-scale LED UV systems was not available).
• It is also assumed that the LED UV lamps are housed in the same stainless steel
vessel with electronic controls as the conventional UV, and that the lifetime of
these components is five years.
• Based on equipment specifications from Aquionics, 0.0039 kWh of electricity are
required per m3 water treated via LED UV.
• LED lamp infrastructure requirements were modeled based on a U.S. Department
of Energy (DOE) LCA on energy and environmental impacts of LED lighting
53
-------
90
products. This study identified the background ecoinvent data sets and
associated quantities utilized in the life cycle inventory model, and EPA
replicated this LCI model (See Table 5-3, 5-6, and 5-8 of DOE study). The DOE
study (and therefore this study) assumes the LEDs are produced in China.
• Disinfection with LED UV does not result in any formation of DBFs.
• Disinfection with LED UV can lead to the same levels of cryptosporidium
reduction as disinfection with gaseous chlorine.
• No chlorine is required for primary disinfection, but some gaseous chlorine is still
required to maintain a chlorine residual in the distribution system. This is based
on information from FTT (see Section 7.4). The sodium hypochlorite added
during distribution is still added in the same amount as this is required to boost
chlorine residual in certain parts of the distribution system. This sodium
hypochlorite boost may not be applicable for other drinking water systems.
7.3.2 Unit Processes
The specific unit processes added for the LED UV LCA model are identified below.
Disinfection
1. Primary Disinfection, LED UV. Primary disinfection with LED UV. The inputs to this
unit process include operation and infrastructure requirements for the UV units.
2. LED UV Drinking Water Treatment, Operation. This process covers electricity usage
associated with operation of the UV units.
3. LED UV Drinking Water Treatment, Infrastructure. Infrastructure inputs for the UV
units are aggregated in this unit process. Infrastructure processes included are the LED
die fabrication, LED packaging assembly, three-inch sapphire wafer manufacture,
production of the stainless steel UV vessel, and production of the LED UV electronics
control unit.
4. Three-Inch Sapphire Wafer Manufacture. Preparation of sapphire wafers to use for
LED die fabrication.
5. LED Die Fabrication. LED semiconductor device fabrication.
6. LED Packaging Assembly. Packaging and assembly of the LED devices.
7. Stainless Steel UV Vessel. Production of the stainless steel UV vessel.
8. LED UV Electronics Control Unit. Production requirements for an electronics control
unit.
Use
9. Drinking Water Consumption, LED UV. Final delivery of water, which is disinfected
with LED UV, to an average consumer. This unit process aggregates the other main life
cycle stages and is used to build the final product system. There are no actual impacts
associated with the drinking water consumption life cycle stage itself.
20 U.S. Department of Energy: Buildings Technology Program. June 2012. Life Cycle Assessment of Energy and
Environmental Impacts of LED Lighting Impacts. Accessed at:
http ://apps 1. eere. energy .gov/buildings/publications/pdfs/ssl/2012_led_lca-pt2 .pdf
54
-------
Table 20 displays the data sources used for the LED UV model in addition to the data sources
used in the base case model (See Table 3). Aquionics' equipment specifications were used to
determine operational energy requirements. Upstream infrastructure was modeled based on a
DOE LCA of LEDs.20 This study identified the background ecoinvent data sets and associated
quantities utilized in the LCI and EPA replicated this LCI model.
Table 20. LED UV data sources.
Process
LED UV disinfection operation
Infrastructure for UV lamp
Infrastructure for UV vessel and electronics
Three-Inch Sapphire Wafer Manufacture
LED Die Fabrication
LED Packaging Assembly
Materials for LED production
Energy for LED production
Data Source
Aquionics' equipment
specifications
DOE LED LCA2U
Equivalent to conventional UV analysis
DOE LED LCA/U
DOE LED LCA2U
DOE LED LCA2U
ecoinvent v2.2
ecoinvent v2.2
7.3.3 Results
Table 21 provides results of the LED UV analysis on the basis of 1 m3 water delivered to the
consumer. Figure 25 presents comparative summary results by life cycle stage for Base Case 1
versus LED UV disinfection. As previously mentioned, no cost data was available for LED UV
disinfection, so this is excluded from the figure. Figure 26 presents the percent change across the
impact results when using LED UV disinfection rather than gaseous chlorine disinfection.
Section 7.6 presents results comparisons between the three alternative disinfection technologies
and the base case. Overall LED UV results are similar to conventional UV results, but LED UV
is more energy efficient compared to conventional UV. With the exception of the decrease in
hazardous material usage, decrease in human health noncancer impacts, and the elimination of
the formation of DBFs under the LED UV scenario, the LCA results for the gaseous chlorine
base case and primary disinfection with LED UV are essentially equivalent. Human health
noncancer results decrease because of the elimination of gaseous chlorine with the LED UV
disinfection system. The primary emission leading to human health noncancer impacts during the
production of gaseous chlorine is CFC-10.
55
-------
Table 21. LED UV results per m3 drinking water delivered to the consumer.
Results Category
Cryptosporidium
TTHM
Hazardous Materials
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer, Total
Human Health, NonCancer, Total
Human Health, Criteria
Ecotoxicity, Total
Unit
oocyst
kg TTHM
kgCl2
kg CO2 eq
MJ
kg oil eq
kg H+ mole eq
kgN eq
m3
kg O3 eq
kgCFC-lleq
kg Fe eq
CTU
CTU
kgPMWeq
CTU
LEDUV
1.00
0
4.4E-04
1.04
19.8
.36
0.48
9. 7E-04
1.20
0.067
2.8E-08
0.036
2.9E-11
3.0E-11
0.0015
4.5E-04
56
-------
I Source Water Acquisition
Distribution
l Primary Disinfection
I Drinking Water Treatment Plant Energy Usage
Pre-Disinfection
Percent of Maximum
100%
Figure 25. Base Case 1 and LED UV comparative summary results.
57
-------
-80% -70% -60% -50%
Percent Change
-40% -30% -20%
-10%
10%
Hazardous Materials
Human Health, NonCancer
Human Health, Cancer
Ozone Depletion
Blue Water Use
Eutrophication
Energy Demand
Global Warming
Fossil Depletion
Human Health, Criteria
Smog
Acidification
Metal Depletion
Ecotoxicity
Figure 26. Percent change in impacts if using LED UV rather than gaseous chlorine for
disinfection.
Similar to the conventional UV analysis, for most impacts examined (excluding the formation of
DBFs), the change seen for utilization of LED UV versus gaseous chlorine for primary
disinfection is due to increased electricity usage with LED UV. Therefore, a sensitivity analysis
is run here varying the electricity usage for LED UV disinfection +/- 25 percent. Figure 27
presents the results of this analysis. Overall, the total life cycle impacts for DWT disinfection
with LED UV do not vary more than +/- 0.1% when varying the electricity usage for LED UV
operation +/- 25 percent. This is primarily a result of the overall small impact of the primary
disinfection life cycle stage as compared to other life cycle stages that are larger consumers of
energy (e.g., pumping at the DWT plant, distribution of the treated water to the consumer). The
change for the +/- 25 percent electricity usage for LED UV operation is less than that seen in the
same sensitivity analysis conducted for the conventional UV technology, as LED UV requires
less electricity overall for disinfection compared to the conventional mercury-vapor UV.
58
-------
Primary Disinfection
-0.10%
•g .g Min (-25%)
^ g
5 | Max (+25%)
Min (-25%)
t9 Q Max (+25%)
Min (-25%)
o ~
fe ,§" Max (+25%)
« c
^3 o
Min (-25%)
•Jj Max (+25%)
Min (-25%)
S "
Max (+25%)
Min (-25%)
Max (+25%)
-------
7.4 Ferrate Technology
FTT patented an on-site reactor for municipal and industrial water treatment applications (the
Ferrator®). The Ferrator® generates ferrate ions (FeC>42~) on-site from caustic, sodium
hypochlorite, and ferric chloride and delivers it continuously to the process. Ferrate can be used
as an oxidant, coagulant, and disinfectant. When used at the beginning of a treatment train,
ferrate will oxidize organics and sulfides, eliminate taste and odor issues, and eliminate the need
for GAC to remove disinfection byproducts. According to FTT, ferrate has the following benefits
over chlorine disinfection:
• Reduces the chlorine dose required to maintain an adequate residual;
• Eliminates the need for alum coagulation;
• Reduces the amount of sludge generated; and
• Eliminates the generation of DBFs.
According to FTT, one of the key benefits is the reduction of DBFs. In conventional drinking
water plants, DBFs form when chlorine reacts with the organics present in the raw water. Using
ferrate at the pre-disinfection stage can remove solids and organics. The Ferrator® reactor
controls the generation of ferrate ions such that the chlorine in the chemical feedstocks is
consumed in the reaction to form sodium chloride, which will not combine with organics to form
DBFs. Ferrate also provides disinfection by inactivating microorganisms. Chlorine will still need
to be added to maintain a chlorine residual in the distribution system; however, the quantity of
chlorine required for the residual is reduced and the chlorine is added after all organics have been
removed, eliminating the formation of DBFs.
7.4.1 Data Collection and System Boundaries
EPA obtained information on the Ferrator® technology directly from FTT. Based on discussions
with FTT, EPA made the following changes to the base case model to represent use of the
Ferrator® technology:
• Added 3 ppm ferrate at the pre-disinfection stage as an oxidant/coagulant and eliminated
the addition of alum and polymer as coagulants.
• Added ferric chloride for pH adjustment after addition of ferrate because ferrate will
increase the pH (0.075 ppm of 40% concentration ferric chloride used at the
sedimentation stage and 0.05 ppm of 40% concentration ferric chloride used at the
conditioning stage). Eliminated the addition of lime and sodium hydroxide for pH
control.
• Removed GAC. EPA assumed that ferrate would oxidize any organics present in the raw
water and eliminate any taste and odor concerns; therefore, GAC is not required.
• Reduced volume of sludge generated (see details below).
• Added 2 ppm ferrate for primary disinfection.
60
-------
• Reduced amount of chlorine required by 75 percent. This chlorine dose is required to
maintain a chlorine residual in the distribution system.
• Increased electricity consumption to power the Ferrators®.
• Added infrastructure requirements for production of the Ferrator® (amortized over the
useful life of the system).
EPA compared data provided by FTT to the base case model assumptions and made some
adjustments to the data inputs as described below:
1. Ferrator chemical feedstocks - The chemical composition of Ferrate is confidential. For
purposes of this analysis, EPA assumed ferrate is produced on-site at the DWT plant in a
Ferrator® using sodium hydroxide (50% concentration), sodium hypochlorite (15%
concentration, estimated based on available data), and ferric chloride (40% concentration)
at a mass ratio of 3:1:0.5.
2. Sludge generation - GCWW does not dewater sludge from the sedimentation basin and
returns a watery sludge stream to the river. EPA computed an estimated mass of sludge
generated given the TSS concentration of raw river water and GCWW's dosage of alum,
polymer, and lime, which all contribute to the sludge generated. EPA calculated
approximately 3.6 Ib dry sludge is generated per 1,000 ft3 of water produced in the base
case given a raw water TSS concentration of 43 mg/L.21 EPA performed similar
calculations to determine the amount of sludge generated from ferrate (3.0 Ib dry sludge/
1,000 ft3 of water).
Additional assumptions specific to the LCA model and cost analysis are described in the
subsections below. Because use of ferrate impacts or eliminates the need for many unit processes
in the base case, the system boundaries for the ferrate drinking water treatment model are
provided in Figure 28.
21 Calculations based on equations from Appendix E Sludge Production from Coagulants and Other Treatment
Chemicals, AWWA Research Foundation, "Trace Contaminants in Drinking Water Chemicals", 2002.
61
-------
LEGEND
^~ ~""\
i
0
Primary Input/
Final Demand
Primary Process
Reference
Supply Chain
Or gate (multiple
outputs)
\ "x Or gate (multiple
/ -S inputs)
Sodium
Hexametaphosphate
Production
Sodium
Hypochlorite
Production
1
Distribution
1
Transport, Treated
Drinking Water, Water
Supply Pipeline
*Gaseous chlorine is not used as primary disinfectant, and is added to
maintain a chlorine residual in the distribution system.
/ Drinking Water \
\ Consumption I
V ^
Figure 28. System boundaries of ferrate drinking water treatment.
7.4.2 LCA Model
This section provides information on the unit processes developed and data sources used for the
ferrate DWT LCA model.
7.4.2.1 Unit Processes
EPA developed new unit processes for the specific ferrate DWT processes listed below
(categorized by the overall life cycle stage or material). As shown in Figure 28, the ferrate DWT
unit processes start with source water acquisition and end with drinking water use. Unaffected
62
-------
unit processes from the base case are not listed here. Additional ferrate unit processes from
background LCI database (e.g., ecoinvent v2.2 and U.S. LCI) that have not been modified are
identified in Section 7.4.2.2, Table 22 (Ferrate Data Sources).
Ferrate Solution
1. Ferrate Solution. Ferrate solution is produced on-site at the DWT plant in a Ferrator®
using sodium hydroxide (50%), sodium hypochlorite (15%, estimated based on available
data), and ferric chloride (40%) at a mass ratio of 3:1:0.5. Some electricity is required for
operating the Ferrator® to produce the ferrate solution at the DWT plant. The Ferrator®
infrastructure, amortized over lifetime production of ferrate, is an input to the ferrate
solution unit process.
2. Ferrator® (FeSOO) Production. This unit process includes production of one Fe300
Ferrator®. The Ferrator® has a lifetime of 15 years, weighs 14,000 Ibs, and is composed
primarily of steel (for the frame and skid) and PVC (for piping valves and fittings). EPA
assumed the Ferrator® is, by weight, 90% steel and 10% PVC. This unit process only
includes material production for the Ferrator®, no assembly information was available,
and is therefore excluded.
P re-Disinfection
3. Pre-Treatment, Ferrate. Ferrate solution is used at 3 ppm during pre-treatment to act as
an oxidant/coagulant.
4. Sedimentation, Ferrate. Ferric chloride is added during sedimentation to adjust pH (as
opposed to lime addition in the base case). The sedimentation unit process has been
otherwise unchanged from the base case.
5. Disposal, Sedimentation Waste, Ferrate. Use of ferrate decreases the suspended solids
and total sludge amount since alum is no longer used as a coagulant. Waterborne
emissions of aluminum from the base case are also removed with the elimination of
flocculation in the ferrate model. The overall waterborne emissions of ammonia, BOD,
and COD remain unchanged from the base case, since it is assumed ferrate removes these
emissions at the same rate as the base case model.
6. Conditioning, Ferrate. Ferric chloride is used for pH adjustment, and the use of sodium
hydroxide is eliminated.
7. Pre-Disinfection, Ferrate. This unit process aggregates the upstream ferrate pre-
disinfection unit processes from ferrate pre-treatment through conditioning.
Disinfection
8. Primary Disinfection, Ferrate. Representative of a DWT system using ferrate solution
for primary disinfection. Ferrate is used at a 2 ppm dosage for primary disinfection. Some
gaseous chlorine (reduced 75% from base case) is still included to have a chlorine
residual in the distribution system.
Use
9. Drinking Water Consumption, Ferrate. Final delivery of water to an average
consumer. This unit process aggregates the other main ferrate life cycle stages and is used
to build the final product system. There are no actual impacts associated with the
drinking water consumption life cycle stage itself.
63
-------
7.4.2.2 Data Sources
Table 22 displays the data sources used for the ferrate model in addition to the data sources used
in the base case model (See Table 3). In general, data from FTT were used where available. For
upstream processes that would not be known by FTT such as information on chemical feedstock
production (e.g., ferric chloride, sodium hydroxide, and sodium hypochlorite), EPA used
information from either the U.S. LCI Database or ecoinvent v2.2. Data sets from U.S. LCI
Database and ecoinvent v2.2 have not been adapted for this project.
Table 22. Ferrate data sources.
Process
Ferrate Solution
Sodium Hypochlorite
Ferric Chloride
Sodium Hydroxide
Ferrator® Infrastructure
Polyvinyl Chloride Resin
Steel, Low-Alloyed
Pre-Treatment, Ferrate
Primary Disinfection, Ferrate
Data Source
Data Collection-FTT
ecoinvent v2.2
ecoinvent v2.2
ecoinvent v2.2
Data Collection-FTT
U.S. LCI
ecoinvent v2.2
Data Collection-FTT
Data Collection-FTT
7.4.3 Cost Analysis
„,„, 99
Table 23 lists information FTT provided for use in the cost analysis. After adding in the
Ferrator® system (including electricity usage, chemical inputs, incidental repairs, and amortized
capital investment); reducing the gaseous chlorine usage; eliminating GAC, alum, polymer, lime,
and caustic soda; and reducing the sludge produced, the total annual cost is $8,333,000, a
decrease of $1,659,000 from Base Case 1.
Table 23. Cost data provided by FTT.
Cost Element
Ferrator®
Ferrator® monitor
Ferrator® lifetime
Electricity requirement
Incidental repairs
Value
$810,000a
$20,000
15
15,208b
2%c
Unit
$/unit
$/unit
years
kWh/unit
% of capital cost
Ferrator® units
of
aFTT noted that quantity discounts Ferrator® costs.
bFTT estimated 91,250 kWh of electricity would be required to operate six Ferrators®, which is
approximately 15,208 kWh per Ferrator®.
0 FTT estimated that incidental repairs would cost approximately 2% of the total cost of Ferrator® units
installed, which is approximately $97,200 for six Ferrators®.
: Input data, calculations, and results for the FTT cost analysis are included in the supporting Excel file.
64
-------
In addition, EPA included the following information and assumptions based on discussions with
FIT:
1. The number of Ferrators® required depends on the ferrate dose needed to achieve the
treatment objectives. EPA assumed ferrate would be added at the pre-disinfection stage
as an oxidant/coagulant at a dose of 3 ppm and as the primary disinfectant at a dose of 2
ppm. FIT estimated that this dose would require 5 Ferrators® for pre-treatment and 4 for
disinfection based on a DWT plant capacity of 100 MGD. FTT also recommended
including an additional Ferrator® and an additional Ferrator® monitor as back-ups. EPA
scaled-up the estimates provided by FTT to match the actual volume of water treated by
GCWW used in the base case model. EPA estimated that 10 active Ferrators® and one
active monitor would be required with an additional Ferrator® and monitor as back-ups.
The resulting total capital equipment cost is $8,950,000.
2. Costs include two Ferrate monitors (primary and backup). The required dose of Ferrate is
generated on site. The monitors adjust the ferrate dose to match demand automatically.
The monitor continuously measures and records the concentration of ferrate in the stream
being treated after the ferrate is mixed.
3. Cost multipliers are often applied to equipment costs to account for other direct costs
such as installation, site work, and ancillary equipment and indirect costs such as
permitting, monitoring, and training. Ferrators® are a pre-assembled skid-mounted
system that can be set up on a pad. FTT noted that values less than standard costs
multipliers would be appropriate for estimates of other direct costs and indirect costs. A
2008 AWWA drinking water report used the following multipliers to develop costs for
drinking water residuals processes:
• Piping and fittings - 10% of equipment
• Electrical - 15% of equipment, piping
• Instrumentation - 15% of equipment, piping, electrical
• Contingency, bonding, and mobilization - 25% of total equipment, piping, electrical,
and instrumentation.2
FTT noted there is little ancillary equipment required other than feedstock storage tanks
and transfer pumps. Ferrators® are also self-contained on their own skid and only require
connections to utilities and feedstock tanks. FTT usually connects piping as part of their
contract, so only power connections are required. Because FTT provides the required
instrumentation and controls, EPA only added a 25% cost factor to the capital costs of the
Ferrators® to account for any contingencies. The resulting total capital investment is
$11,177,500.
4. EPA amortized the total capital costs over the 15-year expected lifetime of a Ferrator®
using a bond rate of 6 percent. The resulting annual, amortized cost is $1,151,000.
23 AWWA, 2008. Costing Analysis to Support National Drinking Water Treatment Plant Residuals
Management Regulatory Options. Submitted by Environmental Engineering & Technology, Inc.
Newport News, VA.
65
-------
5. EPA did not include cost credits for any equipment that is no longer required with use of
the Ferrators®. Plants would no longer need GAC equipment and may be able to reduce
equipment required for chlorine addition because less chlorine is required. EPA did
include the annual operating cost reduction from the reduced chlorine use and elimination
of GAC.
7.4.4 Results
Table 24 provides results of the ferrate analysis on the basis of 1 m3 water delivered to the
consumer. Figure 29 displays the ferrate results compared to Base Case 1 and Base Case 2
results by life cycle stage. As can be seen in the figure, only the impacts associated with pre-
disinfection and primary disinfection (green and purple bars) change when switching from Base
Case 1 to the ferrate DWT system. Figure 30 shows the percent change by impact when using
ferrate for pre-treatment and primary disinfection instead of the Base Case 1 scenario. Results
are sorted in this figure to visually display which impact categories are most affected by use of
ferrate.
Ferrate findings of note include:
• Cost results decrease 18 percent when switching from Base Case 1 to the ferrate DWT
system. While primary disinfection costs increase due to the ferrate infrastructure, these
costs are offset and savings are realized by cost reductions in the pre-disinfection stage.
Ferrate cost savings are dominated by: 1) elimination of GAC replacement, 2)
elimination of alum coagulant for flocculation, 3) elimination of sodium hydroxide for
pH adjustment, and 4) elimination of natural gas combustion for regeneration of the
GAC.
• Usage of gaseous chlorine for primary disinfection decreases 75 percent when using
ferrate to maintain a chlorine residual in the distribution system. The sodium hypochlorite
added during distribution is still added in the same amount as this is required to boost
chlorine residual in certain parts of the distribution system. This sodium hypochlorite
boost may not be applicable for other drinking water systems. As discussed previously,
using ferrate at the pre-disinfection stage can remove solids and organics. Since chlorine
is only added after the organics have been removed, DBFs are not expected to form from
using ferrate as applied in this model.
• There is no expected change in human exposure to cryptosporidium when switching to a
ferrate treatment system.
• Global warming potential decreases seven percent when using ferrate compared to the
base case. This reduction is largely attributable to the removal of sodium hydroxide and
lime for pH adjustment in the sedimentation and conditioning processes (ferric chloride is
used for pH adjustment in ferrate model) and the elimination of the GAC adsorption step.
Overall, electricity consumption at the plant and during distribution is the largest
contributor to the GWP. Use of ferrate does not significantly impact electricity usage,
with exception of a small amount of electricity required to operate the Ferrators®. The
66
-------
additional electricity required to operate the Ferrators® has a negligible effect on all
impact results.
• Overall, the additional infrastructure required to produce the Ferrator® units has a
negligible effect on all impact results with the exception of cost.
• Blue water use does not change between Base Case 1 and the ferrate DWT system. Blue
water use is dominated by the actual source water acquired to produce the drinking water
and the water losses during distribution, with neither of these factors being influenced by
the type of disinfection technology.
• Smog formation decreases 10 percent when switching from Base Case 1 to the ferrate
DWT system. This is primarily due to the elimination of the GAC adsorption step, which
includes production of GAC from coal and regeneration of GAC with natural gas, as well
as the elimination of the need for alum coagulant for flocculation since ferrate acts as a
flocculant. Exclusion of the sodium hydroxide and lime for pH adjustment also contribute
to the lower smog results for ferrate. However, a significant decrease in smog formation
is not seen because most of the smog impacts are due to electricity consumption at the
plant and during distribution, which are unaffected by switching to ferrate.
• Similarly, energy demand decreases eight percent and fossil depletion decreases four
percent when switching from Base Case 1 to the ferrate DWT system due to the
elimination of sodium hydroxide, lime, alum coagulant, and GAC adsorption.
• Eutrophication, which is dominated by disposal of the sedimentation sludge, only
decreases one percent under the ferrate DWT system. While elimination of alum
decreases the overall sludge at the DWT plant, it is expected that the same amount of
BOD, COD and ammonia (primary emissions leading to eutrophication) will be removed
from the raw water under the ferrate system; therefore, the final flows of these
waterborne emissions from sedimentation sludge do not vary from the base case.
• Acidification results decrease five percent when switching from Base Case 1 to the
ferrate DWT system due to the elimination of sodium hydroxide (for pH adjustment),
lime, alum coagulant, and GAC adsorption. Acidification impacts in the DWT model are
dominated by sulfur dioxide and nitrogen oxide emissions from fossil fuel combustion for
electricity generation. Again, because ferrate does not influence electricity consumption
significantly at the plant or during distribution, a large overall decrease in acidification
impacts is not realized with the use of ferrate.
• Human health criteria impacts decreases nine percent under the ferrate DWT system.
This is largely due to the reduction in sulfur dioxide emissions with the elimination of
GAC production and regeneration as well as the elimination of the alum coagulant and
sodium hydroxide for pH adjustment.
• Ozone depletion, metal depletion, human health cancer and human health noncancer
results have a higher uncertainty associated with them in the comparative results due to
67
-------
data alignment issues between the unmodified European ecoinvent datasets and the U.S.
datasets (U.S. LCI Database and EPA processes developed for this work). While
reductions in impacts are expected in these categories when switching to ferrate, it is
emphasized that these reductions are likely overstated in the results figures presented.
• Ozone depletion impacts decrease 15 percent under the ferrate DWT system in this
model. This is primarily due to the elimination of sodium hydroxide for pH adjustment
and alum coagulant for flocculation. These materials are modeled using European
ecoinvent datasets. The background European electricity data for production of these
materials uses an electricity grid with higher ozone depletion than the U.S. average
electricity grid modeled for plant operations and drinking water distribution. The average
European electricity grid ozone depletion impacts are primarily influenced by Halon
1301 emissions from crude oil production and Halon 1211 emissions from natural gas
production. These emissions are not incorporated into the U.S. electricity grid fuel
profiles. Therefore, it is expected that the actual reduction in ozone depletion under the
ferrate system is lower than stated here, and the notable reduction is primarily influenced
by data alignment issues. The uncertainty associated with the ozone depletion results is,
therefore, considered high.
• Metal depletion results also decrease 15 percent when switching from Base Case 1 to the
ferrate DWT system. Ecoinvent processes, specifically sodium hydroxide, that are
eliminated with use of ferrate do include background infrastructure for capital equipment.
This metal infrastructure leads to depletion of metals such as nickel, copper and
chromium. So, some reduction in metal depletion is expected when using ferrate;
however, it is likely that the metal depletion reduction value is overstated here. The
ecoinvent data sets and the ferrate production do include background infrastructure, but
background infrastructure is not included for any of the primary DWT processes, U.S.
electricity generation, or background U.S. LCI processes. The uncertainty associated with
metal depletion results is considered high due to these infrastructure data alignment
concerns.
• Human health noncancer results decrease 36 percent under the ferrate DWT system. This
decrease is due to elimination of the sodium hydroxide for pH adjustment, relating to the
background carbon disulfide emissions from ecoinvent European electricity.
• Human health cancer decrease 11 percent when switching from Base Case 1 to the ferrate
DWT system, largely from the elimination of sodium hydroxide for pH adjustment and
alum coagulant for flocculation. This is primarily due to fewer dioxin and formaldehyde
emissions in the background European ecoinvent electricity required to produce these
material.
• Ecotoxicity results decrease 12 percent when switching from Base Case 1 to the ferrate
DWT system. This reduction is due to the elimination of sodium hydroxide for pH
adjustment and alum coagulant for flocculation. The main emissions associated with the
supply chain of these materials that lead to ecotoxicity are cyanide, carbofuran, and
phenol.
68
-------
Table 24. Ferrate results per m3 drinking water delivered to the consumer.
Results Category
Cost
Cryptosporidium
TTHM
Hazardous Materials
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer, Total
Human Health, NonCancer, Total
Human Health, Criteria
Ecotoxicity, Total
Unit
$
oocyst
kg TTHM
kgCl2
kg CO2 eq
MJ
kg oil eq
kg H+ mole eq
kgN eq
m3
kg O3 eq
kgCFC-lleq
kg Fe eq
CTU
CTU
kgPMWeq
CTU
Ferrate
$0.067
LOO
0
4.4E-04
0.97
18.3
0.33
0.45
9.4E-04
1.20
0.060
2.3E-08
0.03 1
2.6E-11
2.0E-11
0.0013
3.9E-04
69
-------
I Source Water Acquisition
Distribution
Pre-Disinfection
I Drinking Water Treatment Plant Energy Usage
Overhead
l Primary Disinfection
0%
20%
Percent of Maximum
40% 60%
80%
100%
+- Base Case 1
o
o .3
I1S
0 o,
1
a
-------
Percent Change
-100% -90% -80% -70% -60% -50% -40% -30% -20% -10% 0%
TTHM
Chlorine Usage
Human Health, NonCancer
Cost
Ozone Depletion
Metal Depletion
Ecotoxicity
Human Health, Criteria
Smog
Human Health, Cancer
Energy Demand
Global Warming
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Cryptosporidium
*Higher uncertainty associated with results for these impact categories. Reductions likely
overstated due to data alignment concerns.
Figure 30. Percent reduction when switching from Base Case 1 to ferrate DWT system.
The ferrate base case assumes that 3 ppm of ferrate is added at the pre-disinfection life cycle
stage and 2 ppm of ferrate is added during the primary disinfection life cycle stage. The actual
ferrate dosage may vary depending on the specific plant conditions and the quality of the
incoming water. A sensitivity analysis is conducted here varying the ferrate dosage during the
pre-disinfection and primary disinfection stages. A minimum dosage of 1 ppm during pre-
disinfection and 1 ppm during primary disinfection and a maximum dosage of 5 ppm during pre-
disinfection and 3 ppm during primary disinfection are investigated.
Figure 31 presents the results of this sensitivity analysis. Impact assessment results do not vary
more than +/- 0.80 percent in this sensitivity analysis; therefore, the ferrate LCA model is not
sensitive to the ferrate dosage requirements.
Figure 32 presents a tornado chart that displays the results of the total cost sensitivity analysis.
The cost sensitivity analysis performed a Monte Carlo simulation, varying the following:
• Pre-disinfection ferrate dose from 1 ppm to 5 ppm (same as was performed for the LCA
sensitivity analysis).
• Disinfection ferrate dose from 1 ppm to 3 ppm (same as was performed for the LCA
sensitivity analysis).
• Cost of a Ferrator® unit by ±10% (baseline value of $810,000/unit).
71
-------
• Bond rate from four to eight percent (a ±33% change from the baseline value of six
percent).
The sensitivity results show that the total costs are most sensitive to the bond rate and followed
by the Ferrator® cost per unit, pre-disinfection ferrate dose, and the disinfection ferrate dose.
However, the total cost changes are within approximately ±1.70%. Therefore, although the total
costs are more sensitive to the bond rate than they are to the capital equipment cost and the
ferrate doses, the total cost sensitivities are mitigated over the parameter values studied. This
result is expected as the Ferrator® system only impacts the pre-disinfection and disinfection
stages, which constitute 5% and 18% of the total costs, respectively (the use of ferrate increases
the disinfection costs but decreases the pre-disinfection costs for an overall cost savings). The
total costs are dominated by the distribution costs (35%) and overhead costs (27%). The plant
energy costs constitute the remaining 15% of the total costs.
It is important to note that FTT has reported to be continuing the optimization of its ferrate
manufacturing equipment, thus reducing the associated equipment costs. Compared with the
estimated costs in this study, a significantly lower cost may occur in the present and future,
especially for large water treatment plants.
72
-------
Percent Change
-0.80% -0.60% -0.40% -0.20% 0.00% 0.20% 0.40% 0.60% 0.80%
•a .s Min
•^ c
^ ^ Max
>^13 Min
£? §
W Q Max
— .2 Min
It
^ Q Max
o
3 Min
o
•5 Max
.§ Min
•S a
5 Max
w
a Min
5 Max
m
BO Min
O
1X1 Max
-------
Relative Change in Total Costs for Ferrator Scenario (%)
-2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00
Bond Rate (%)
Ferrator Cost per Unit ($)
Pre-Treatment Ferrate Dose (ppm)
Disinfection Ferrate Dose (ppm)
Figure 32. Tornado chart of the sensitivity results for the relative changes in total costs for the
ferrate scenario.
7.5 Imaging Systems Technology
1ST is an electronics and materials firm that manufactures a hollow gas encapsulating shell called
a plasma-bead. When a voltage is applied across the shell, the gas ionizes into plasma to generate
UV light which can be used for disinfection. Multiple plasma-beads can be configured in an
array to disinfect different quantities of drinking water. 1ST is at the design phase of
implementing their plasma-bead technology for drinking water disinfection and is looking to
partner with UV vendors such as Aquionics to develop pilot- and full-scale plasma-bead
disinfection technologies. As such, 1ST is not able to provide detailed unit process and cost data
to use to develop a model to compare to the base case. EPA is working with 1ST to develop
general assumptions regarding the manufacturing and composition of their plasma-bead
technology to use in the analysis. According to 1ST, potential benefits of the technology include:
• Low manufacturing cost;
• Low operating cost;
• UV light source is in direct contact with water and the light output is very bright;
• Technology can scale to large sizes; and
• Plasma-bead are composed primarily of alumina oxide gas and do not contain
environmentally hazardous materials such as mercury.
7.6 Comparative Results
Figure 33 presents the summary comparative results of the base case DWT model versus the
alternative disinfection technology models (ferrate, conventional UV, and LED UV). Utilization
of ferrate results in environmental, human health, and cost benefits for combined use in the pre-
disinfection and primary disinfection stages, since ferrate acts as both a coagulant and
disinfectant and only small dosages are required for treatment. Application of UV technology
increases impacts during disinfection through increased electricity consumption and through new
capital investment, but eliminates the formation of disinfection by-products and greatly reduces
74
-------
hazardous chlorine usage. LED UV is more energy efficient compared to conventional mercury-
vapor UV; however, it is currently developed only for point-of-use applications, and not large-
scale treatment facilities.
Figure 34 presents the comparative normalized results for the different disinfection technology
life cycles. The following results are shown on this figure:
• Cost: this category displays cost by life cycle stage. The costs are shown as a percentage
of the highest cost system (in this case conventional UV).
• Normalized by impact: this category presents the normalized impact assessment results
by impact category. Impact categories have been normalized using TRACT v2.1
normalization factors.24 The results are shown as a percentage of the highest normalized
impact system (in this case conventional UV).
• Normalized & weighted by impact: this category presents the normalized and weighted
impact assessment results by impact category. Impact categories have been normalized
using TRACT v2.1 normalization factors and have been weighted using NIST weighting
factors.24, 25 The results are shown as a percentage of the highest normalized impact
system (in this case conventional UV).
• Normalized by stage: this category presents the normalized impact assessment results by
life cycle stage. Life cycle stages have been normalized using TRACT v2.1 normalization
factors.24 The results are shown as a percentage of the highest normalized impact system
(in this case conventional UV).
Only impacts with TRACT normalization factors are shown in Figure 34. Blue water use, metal
depletion, cumulative energy demand, and fossil depletion are excluded due to lack of available
normalization factors. Additional water treatment metrics included (TTHM and hazardous
materials) are not shown since they also do not have associated normalization factors. Cost
results for LED UV are also not shown in Figure 34 due to lack of available cost data for this
technology. Some findings of note from Figure 34:
• Weighting increases the relative importance of global warming potential.
• In all cases, conventional UV has the highest overall normalized impact, normalized and
weighted impact, and cost.
• Impact assessment results' correlate with cost results.
24 Ryberg, M, Vieira, M.D.M., Zgola, M, Bare, I, and Rosenbaum, R.K., 2014. Updated US and Canadian
normalization factors for TRACI 2.1. Clean Techn Environ Policy, 16: 329-339.
25 Gloria, T.P., Lippiatt, B.C., and Cooper, J. 2007. Life cycle impact assessment weights to support environmentally
preferable purchasing in the United States. Environ. Sci. Technol, 41, 7551-7557.
75
-------
i Base Case 1
Ferrate
I Conventional UV
ILEDUV
20%
Percent of Maximum
40% 60%
±
80%
100%
Cost*
Cryptosporidium
TTHM
Hazardous Materials
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer
Human Health, NonCancer
Human Health, Criteria
Ecotoxicity
*No cost data available for LED UV
Figure 33. Summary comparative results of alternative disinfection technologies.
76
-------
20%
Percent Contribution
40% 60%
80%
100%
Cost
&
Cost
Normalized by Impact
O
U
Normalized & Weighted by Impact
Normalized by Stage
• Global Warming
• Eutrophication
Human Health. Criteria
Drinking Water Treatment Plant Energy Usage
• Primary Disinfection
• Overhead
• Acidification
• Smog •<•
Other Inroads**
• Pre-Disinfection
Distribution ^
Impact Categories
Life Cycle Stage
*No cost data available for LED UV
**"Other Impacts" includes human health noncancer, human health cancer, ozone depletion, and
ecotoxicity
Figure 34. Normalized comparative results for different drinking water treatment disinfection
technologies.
77
-------
8. PoiNT-OF-UsE ALTERNATIVE DISINFECTION TECHNOLOGIES
EPA investigated the impacts and costs associated with point-of-use drinking water technologies.
EPA focused this analysis on point-of-use technologies that may be used by hospitals to reduce
pathogen exposure for immune-compromised individuals. EPA did not investigate point-of-use
alternatives for home use. Because point-of-use technologies will not change unit processes at
the drinking water plant, EPA did not compare point-of-use LCA results to the base case model.
Instead, EPA reported the life cycle impacts of each point-of-use technology and compared the
impacts to the additional pathogen removal provided by the technology.
Hospitals draw water from the municipal water supply. Although water is disinfected at the
treatment plant and chlorine is added to maintain an appropriate residual throughout the
distribution system, microorganisms can be present in water at the tap due to residual bacteria in
the distribution systems. Hospitals may use additional technologies to prevent pathogen
exposure. Typically, Legionella and Pseudomonas bacteria are of greatest concern to hospitals.
Hospitals may use technologies that are implemented for the water system as a whole at the point
water enters the building from the municipality and prior to distribution throughout the facility.
However, EPA's analysis focused on point-of-use filters that could be installed at or near the
faucet.
EPA investigated use of Pall-Aquasafe™ 31-day point-of-use filters for waterborne
microorganisms. According to Pall's website, filters can be used for up to 31 days and use a
double-layer sterilizing grade membrane to reduce Legionella and Pseudomonas and other gram-
negative bacteria.26 The cost per filter ranges from $39 to $79, depending on the volume
purchased by each customer. Since the point-of-use filter is an additional level of drinking water
treatment and does not replace any processes in the base case water treatment scenario, the filter
cost does not change any of the costs associated with water treatment in the base case.
An additional point-of-use technology examined was LED UV. As discussed in Section 7.3,
Aquionics' current LED UV system is for point-of-use applications. Aquionics notes that this
system may be used for stand-alone point of use, healthcare equipment, laboratory research
equipment, and autocalves among other uses. This system is not installed directly on the faucet,
but rather more likely installed in the pipe system right before the faucet.
8.1 System Boundaries
The system boundaries for the point-of-use disinfection technologies are displayed in Figure 35.
Prior to point-of-use disinfection, all processes are equivalent to base case 1. The drinking water
at the hospital then undergoes further disinfection via either the point-of use faucet filter (Pall) or
the LED UV technology (Aquionics). The system boundaries end at consumption of the water by
an immune-compromised adult.
26 Pall Corporation Aquasafe Medical Filters. See: http://www.pall.com/main/medical/product.page?id=45154#
78
-------
f Drinking Water
Consumption by Immune-
Figure 35. System boundaries for hospital point-of-use drinking water treatment.
8.2 Pall Point-of-Use Filter
8.2.1 LCA Model
EPA made the following assumptions for the point-of-use filter analysis:
79
-------
• For the point-of-use LCA model, only the primary disinfection stage is changed
from the base case DWT model.
• Water treated per filter unit is highly dependent upon the water use patterns in a
given hospital. Thus, it is assumed that an individual faucet and filter are used on
average 12 hours per day for one month. The faucet and filter is assumed to flow
at a rate of 8.33 liters of treated water per minute, which is based on the standard
maximum flow for faucets in the U.S. set by the U.S. Department of Energy and
is within the rate of disinfection for the Pall Aquasafe 31-day filter as reported on
97 9R
their website. '
• Point-of-use filter infrastructure requirements were modeled based on the publicly
available Declaration of Compliance for the Aquasafe 31-day filter, specifications
for the QPoint™ filter published on the Pall Corporation website, and personal
communication with Pall representatives.29'30'31 This study identified the
background ecoinvent datasets and associated quantities utilized in the life cycle
inventory model, which were replicated in this LCI model.
• Disinfection with point-of-use filters removes 100% of Legionella and
Pseudomonas present in drinking water delivered to the hospital. This is based on
field evaluation reports on the Pall Aquasafe 31-day filter.32
8.2.2 Unit Processes
The specific unit processes added for the point-of-use filter LCA model are identified below.
Infrastructure
1. Point-of-Use Hospital Filter, Infrastructure. Infrastructure inputs for the point-of-use
hospital filter are aggregated in this unit process. Infrastructure processes included are the
production of the filter itself, production of a tap adapter, and corrugated packaging for
distribution of the filters to hospitals.
2. Point-of-Use Hospital Filter, Production. Filters are manufactured from a variety of
plastic resins.
3 Tap Adapter for Point-of-Use Hospital Filter. The faucet adapter, made of nickel-
plated brass connects the point-of-use filter to a standard faucet for use in a hospital.
4. Packaging for Point-of-Use Hospital Filter. Filters are shipped to hospitals in
corrugated boxes with 12 filers per box.
27 U.S. Department of Energy: Buildings Technology Program. Oct 2013. Faucets. Accessed at:
http://wwwl.eere.energy.go^uildings/appliance_standards/product.aspx/productid/64
28 Pall Corporation. Pall-Aquasafe™ AQ3 IF IS and AQ31F1R Filters for Waterborne Microorganisms. Accessed at:
http://www.pall.com/main/medical/product.page?id=45154
29 Pall Corporation. March 2013. Declaration of Compliance: Pall-Aquasafe™ Disposable Water Filter 31 Day Use
- Tap Application. Accessed at: http://www.pall.com/pdfs/Medical/AQ31FlR-Declaration-of-Compliance.pdf
30 Pall Corporation. Nov 2012. QPoint™ Tap Water Filter - USA. Accessed at:
http://www.pall.com/main/consumer-water/product. page?lid=h8pw!57j
31 Pall Medical North American Sales Representatives Personal Communication. February 24, 2014.
32Pall Corporation. Feb 2009. Pall-Aquasafe™ Disposable Water Filter - Tap (AQ31F1S and AQ31F1R) Field
Evaluation Report.
80
-------
Use
5. Drinking Water Consumption, Base Case, at Hospital with Point-of-Use Filter. The
point-of-use filter removes Legionella and Pseudomonas and other gram-negative
bacteria from drinking water at the tap.
Table 25 displays the data sources used for the point-of-use hospital filter in addition to the data
sources used in the base case model (See Table 3). Some data on components and weight of the
filter were gathered from Pall. For upstream processes that would not be known by Pall such as
information on resin production, EPA used information from the National Renewable Energy
Laboratory's U.S. Life Cycle Inventory Database (U.S. LCI), a publically available life cycle
inventory source.33 Where data were not available from Pall or the U.S. LCI, ecoinvent v2.2,
EPA used a private Swiss LCI database with data for many unit processes.34
Table 25. Point-of-Use hospital filter data sources.
Process
Point-of-use hospital filter production
Corrugated for filter packaging
Nickel-plated brass tap adapter
Polycarbonate for filter
High-density polyethylene resin for filter
Synthetic rubber for filter
Polypropylene for filter
Injection molding of plastic components of
filter
Data Source
Information from Pall
ecoinvent v2.2
ecoinvent v2.2
ecoinvent v2.2
U.S. LCI
ecoinvent v2.2
U.S. LCI
ecoinvent v2.2
8.2.3 Results
Table 26 displays results for the base case and base case plus the point-of-use hospital filter per
cubic meter of drinking water delivered to the immune-compromised person. Figure 36 presents
summary results by life cycle stage for Base Case with the additional point-of-use disinfection
with the Pall Aquasafe 31-day filter. As previously mentioned, no cost data was available for
point-of-use filtration, so this is excluded from the figure. Overall point-of-use filter results show
minimal increases in impacts compared to the base case results.
National Renewable Energy Lab. US LCI Database. See: http://www.nrel.gov/lci/database/default.asp.
34 Ecoinvent Centre (2010), ecoinvent data v2.2. ecoinvent reports No. 1-25, Swiss Centre for Life Cycle
Inventories.
81
-------
Table 26. Base case and Base case plus point-of-use hospital filter results per m3 drinking water
delivered to the consumer.
Results Category
Hazardous Materials
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer, Total
Human Health, NonCancer, Total
Human Health, Criteria
Ecotoxicity, Total
Unit
kgCl2
kg CO2 eq
MJ
kg oil eq
kg H+ mole eq
kg N eq
m3
kg O3 eq
kg CFC-11 eq
kg Fe eq
CTU
CTU
kgPMWeq
CTU
Base Case 1
0.0018
1.04
19.8
0.36
0.48
9. 7E-04
1.20
0.067
2.8E-08
0.036
2.9E-11
3.2E-11
0.0015
4.4E-04
Base Case 1 plus
Point-of-Use
Hospital Filter
0.0018
1.04
19.9
0.36
0.48
9. 7E-04
1.20
0.067
2.8E-08
0.036
2.9E-11
3.2E-11
0.0015
5.9E-04
82
-------
I Source Water Acquisition
] Pre-Disinfection
I Distribution
Point-of-Use Disinfection
I Drinking Water Treatment Plant Energy Usage
I Primary Disinfection
Overhead
0%
Chlorine Usage
Global Warming
Energy Demand |
Fossil Depletion |
Acidification |
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
20%
Percent Contribution
40% 60%
80%
100%
Human Health,
Cancer
Human Health,
NonCancer
Human Health,
Criteria
=
Ecotoxicity
Figure 36. Base Case 1 plus point-of-use hospital filter contribution analysis results.
Because the point-of-use filter results are dependent on the assumption regarding water use per
day (base case assumed 12 hours per day), additional analyses were conducted assuming 1 hour
use per day and 24 hour use per day. The percent change in impacts for the base case plus the
point-of-use hospital filter compared to the base case without the point-of-use hospital filter was
calculated for the three different use scenarios. The results of this analysis are displayed in
Figure 37. Ecotoxicity is excluded, since it has a comparatively large increase and makes it
difficult to interpret other impact changes graphically. Ecotoxicity impacts are largely driven by
upstream fungicide and pesticide use during potato farming for the potato starch in the
corrugated boxes used to distribute the filters. Overall, impacts increase with less water treated
per day, since this means more filters are required per volume of water.
83
-------
I Increase with 1 hr day use • Increase with 12 hr day use • Increase with 24 hr day use
0%
5%
Percent Change
10% 15% 20%
25%
30%
Global Warming
Energy Demand
Fossil Depletion 1
Acidification 1
Eutrophication 1
Blue Water Use
.
Smog
Ozone Depletion
Metal Depletion 1
Human Health, Cancer 1
Human Health, NonCancer
Human Health, Criteria
1
8.3
Figure 37. Base case percent change with point-of-use filter.
LED UV Point-of-Use Filter
8.3.1 LCA Model
EPA made the following assumptions for the point-of-use LED UV analysis:
• The LED UV system modeled is identical to that modeled in Section 7.3, with the
following exceptions.
o Instead of being housed in a stainless steel vessel with electronic controls, it is
assumed the LED lamp is within a 6 pound unit that is primarily polypropylene
with stainless steel pipe attachments.35
The unit is 6 Ib per Aquionics website: http://www.aquionics.com/main/pearl-brand2/pearlaqua/. EPA assumed
the plastic housing was polypropylene due to lack of specific composition data.
84
-------
o Based on Aquionics' website, it is assumed that 1 LED lamp treats 60,000 gallons
of water over its lifetime.36
8.3.2 Unit Processes
The specific unit processes added for the point-of-use LED UV unit LCA model are identified
below.
Disinfection
1. Disinfection, Point-of-Use LED UV. Primary disinfection with LED UV. The inputs to
this unit process include operation and infrastructure requirements for the UV units.
2. Point-of-Use LED UV Drinking Water Treatment, Operation. This process covers
electricity usage associated with operation of the point-of-use UV units.
3. Point-of-Use LED UV Drinking Water Treatment, Infrastructure. Infrastructure
inputs for the UV units are aggregated in this unit process. Infrastructure processes
included are the LED die fabrication, LED packaging assembly, three-inch sapphire
wafer manufacture, and the point-of-use UV vessel.
4. Point-of-Use UV Vessel. Production of the plastic and steel vessel used to house the
LED UV lamps.
Use
5. Drinking Water Consumption, Base Case, at Hospital with Point-of-Use LED UV.
Final delivery of water, which is disinfected with LED UV, to an immune-compromised
adult. This unit process aggregates the other main life cycle stages and is used to build
the final product system. There are no actual impacts associated with the drinking water
consumption life cycle stage itself.
Table 27 displays the data sources used for the point-of-use LED UV model in addition to the
data sources used in the base case model (See Table 3). Aquionics' equipment specifications
were used to determine operational energy requirements. Upstream infrastructure was primarily
modeled based on a DOE LCA of LEDs.20 This study identified the background ecoinvent data
sets and associated quantities utilized in the DOE LCI and EPA replicated this LCI model.
Aquionics' equipment specifications were also used to determine the materials and weights of
the UV vessel.
36 Aquionics. PearlAqua™. Accessed at: http://www.aquionics.com/main/pearl-brand2/pearlaqua/ (February 10.
2014).
85
-------
Table 27. Point-of-Use LED UV data sources.
Process
Point-of-Use LED UV disinfection
operation
Infrastructure for UV lamp
Infrastructure for Point-of-Use UV vessel
Three-Inch Sapphire Wafer Manufacture
LED Die Fabrication
LED Packaging Assembly
Materials for LED production
Energy for LED production
Data Source
Aquionics' equipment specifications
DOE LED LCA2U
Aquionics' equipment specifications
DOE LED LCA2U
DOE LED LCA/U
DOE LED LCA2U
ecoinvent v2.2
ecoinvent v2.2
8.3.3 Results
Table 28 presents results for the base case and base case plus the point-of-use hospital LED UV
system per cubic meter of drinking water delivered to the immune-compromised person. Figure
38 shows summary results by life cycle stage for Base Case with the additional point-of-use
disinfection with the LED UV unit. As previously mentioned, no cost data was available for
LED UV, so this is excluded from the figure. A notable increase in overall impacts is seen for the
addition of point-of-use LED UV disinfection. While some of this increase is due to electricity
requirements for LED UV disinfection, the majority of increased impacts are driven by
production of the LED UV lamps. The LED UV lamp infrastructure (e.g., sapphire wafer
manufacture, die fabrication) is complex, and the lamps are assumed to be produced in China,
which generates much of its electricity from coal, a relatively high impact energy source. The
electricity mix in China is modeled based on ecoinvent v2.2 data specific to China, with 78.6%
of the electricity sourced from hard coal, followed by 15.9% sourced from hydropower, 2.9%
sourced from oil, and 2.1% sourced from nuclear12 Such LED UV infrastructure burdens are not
seen for the large-scale LED UV analysis, as that analysis assumes 200 million gallons of water
is able to be treated per lamp compared to the 60,000 gallons of water treated per lamp in this
point-of-use analysis.
86
-------
Table 28. Base case and Base case plus point-of-use hospital LED UV disinfection results per m3
drinking water delivered to the consumer.
Results Category
Hazardous Materials
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion
Human Health, Cancer, Total
Human Health, NonCancer, Total
Human Health, Criteria
Ecotoxicity, Total
Unit
kgCl2
kg C02 eq
MJ
kg oil eq
kg H+ mole eq
kgN eq
m3
kg O3 eq
kg CFC-11 eq
kg Fe eq
CTU
CTU
kgPMWeq
CTU
Base Case 1
0.0018
1.04
19.8
0.36
0.48
9. 7E-04
1.20
0.067
2.8E-08
0.036
2.9E-11
3.2E-11
0.0015
4.4E-04
Base Case 1 plus
Point-of-Use
Hospital LED UV
Disinfection
0.0018
1.47
25.6
0.48
0.67
1.4E-03
1.21
0.101
4.0E-08
0.075
4.6E-11
5.0E-11
0.0023
1.8E-03
87
-------
I Source Water Acquisition
] Pre-Disinfection
I Distribution
I Drinking Water Treatment Plant Energy Usage
I Primary Disinfection
IPoint-of-Use LED UV Disinfection
20%
Percent Contribution
40% 60%
0%
Chlorine Usage
Global Warming
Energy Demand
Fossil Depletion
Acidification
Eutrophication
Blue Water Use
Smog
Ozone Depletion
Metal Depletion |
Human Health,.. |
Human Health,..
Human Health,..
Ecotoxicity
Figure 38. Base Case 1 plus point-of-use hospital LED UV disinfection contribution analysis results.
8.4 Comparative Results
Figure 39 illustrates the comparative results for the different hospital point-of-use disinfection
technologies. Both point-of-use technologies are examined as an addition to Base Case 1
(disinfection with gaseous chlorine). In this figure, results are normalized to the point of use
technology with the highest impact in the category under examination. In all cases, the LED UV
point-of-use technology has the greater impacts compared to the Pall point-of-use tap filter. The
LED UV system requires some electricity for operation; whereas, the filter does not require
electricity for generation. The production of the LED UV lamp in China is relatively more
burdensome for the impacts examined compared to the infrastructure production requirements of
the Pall filter.
While a direct comparison is made here between these two point-of-use disinfection
technologies, there are some key distinctions between them. The Pall filter is designed for
100%
88
-------
application to the faucet; whereas, the LED UV system is designed for application prior to the
faucet. Any pathogens formed near the faucet may not be treated by the LED UV system.
Additionally, the Pall filter is designed specifically for hospital use; whereas, Aquionics notes
that healthcare is just one of many applications for the point-of-use LED UV system. This
analysis is provided to begin to understand the potential impact differences between these two
systems, and it is not intended to provide a recommendation on use of either of the technologies.
89
-------
l Source Water Acquisition
Pre-Disinfection
Distribution
l Drinking Water Treatment Plant Energy Usage
l Primary Disinfection
Point-of-Use Disinfection
20%
Percent Contribution
40% 60%
80%
100%
5
Point-of-Use Filter
Point-of-Use LED UV
Point-of-Use Filter
Point-of-Use LED UV
gjj "B Point-of-Use Filter
S3 g
W Q Point-of-Use LED UV
-.2 Point-of-Use Filter
$ "S
r9 "3-
Point-of-Use LED UV
'^3
i fe
o
<
Point-of-Use Filter
Point-of-Use LED UV
tro
tio
Point-of-Use Filter
Point-of-Use LED UV
Point-of-Use Filter
U
5 Point-of-Use LED UV
03
OD Point-of-Use Filter
Point-of-Use LED UV
•o .o Point-of-Use Filter
Point-of-Use LED UV
Point-of-Use Filter
Point-of-Use LED UV
j= 3
~
Point-of-Use Filter
O Point-of-Use LED UV
Point-of-Use LED UV
Point-of-Use Filter
S "3 £
^ 3 Point-of-Use LED UV
1
Point-of-Use Filter
Point-of-Use LED UV
Figure 39. Comparative results for different hospital point-of-use disinfection technologies.
90
-------
9. OVERALL RESULTS SUMMARY
Results of the base case drinking water analysis with disinfection via gaseous chlorine show
impacts are largely driven by electricity consumption at the drinking water treatment plant and
during distribution to the consumer. Overall, primary disinfection with gaseous chlorine only
contributes zero to five percent to the total life cycle impacts of drinking water treatment for the
results categories examined. Utilization of ferrate results in environmental, human health, and
cost benefits for combined use in the pre-disinfection and primary disinfection stages, since
ferrate acts as both a coagulant and disinfectant and only small dosages are required for
treatment. Application of UV technology increases impacts during disinfection through increased
electricity consumption and through new capital investment, but eliminates the formation of
disinfection by-products and greatly reduces hazardous chlorine usage. LED UV is more energy
efficient compared to conventional mercury-vapor UV; however, it is currently developed only
for point-of-use applications, and not large-scale treatment facilities. For hospital point-of-use
disinfection, the LED UV technology has the greater impacts overall compared to the Pall filter.
The LED UV system requires some electricity for operation; whereas, the filter does not require
electricity for generation and the production of the LED UV lamp in China is relatively more
burdensome for the impacts examined compared to the infrastructure production requirements of
the Pall filter. In general, this analysis is provided to understand the potential impacts and trade-
offs between different drinking water disinfection technologies within the framework of the
entire drinking water supply system, and it is not intended to provide a recommendation on
whether any technology is superior to other technologies. The LCA model and cost analysis built
here can serve as the basis for future assessments of water-related technologies and can be
incorporated into broader, sustainable systems analyses of water technologies.
10. REFERENCES
1. Aquionics. Pearl Aqua™, http://www.aquionics.com/main/pearl-brand2/pearlaqua/
(Accessed February 10. 2014).
2. AWWA Research Foundation. 2002. Trace Contaminants in Drinking Water Chemicals.
3. AWWA. 2008. Costing Analysis to Support National Drinking Water Treatment Plant
Residuals Management Regulatory Options. Submitted by Environmental Engineering &
Technology, Inc. Newport News, VA.
4. Chowdhury, Zaid K.; Westerhoff, Garret P; Summers, R. Scott; Leto, Brian; Nowack,
Kirk. 2012. Activated Carbon: Solutions for Improving Water Quality. American Water
Works Association.
5. Ecoinvent Centre (2010), ecoinvent data v2.2. ecoinvent reports No. 1-25, Swiss Centre
for Life Cycle Inventories.
6. Ecoinvent Cumulative Energy Demand (CED) Method implemented in ecoinvent data
v2.2. 2010. Swiss Centre for Life Cycle Inventories.
91
-------
7. eGRID 2008 (Emissions and Generation Resource Integrated Database). U.S. EPA.
(www.epa.gov/cleanenergy/egrid).
8. Ekwall, Cecilia. 2004. LCA of tap water disinfection - a comparison of chlorine and UV-
light. Department of Biometry and Engineering, Swedish University of Agricultural
Sciences. http://ex-epsilon.slu.se/archive/00000280/01/cecilia_ekwall_0402.pdf
9. Goedkoop. M.J., Heijungs. R, Huijbregts. M., De Schryver. A., Struijs. J., Van Zelm. R.
ReCiPe 2008. A life cycle impact assessment method which comprises harmonised
category indicators at the midpoint and the endpoint level; First edition Report I:
Characterisation; 6 January 2009. See http://www.lcia-recipe.net.
10. Gloria, T.P., Lippiatt, B.C., and Cooper, J. 2007. Life cycle impact assessment weights to
support environmentally preferable purchasing in the United States. Environ. Sci.
Technol, 41,7551-7557.
11. Malley, J.P., Jr. 2006. Development of Standard Operating Plans for Mercury Release
from UV Technologies. Used in Drinking Water Treatment Plants. Course Lecture
Materials University of New Hampshire, Durham, NH
12. National Renewable Energy Lab. US LCI Database. See:
http://www.nrel.gov/lci/database/default.asp.
13. Pall Corporation. Pall-Aquasafe™ AQ3 IF 1S and AQ3 IF 1R Filters for Waterborne
Microorganisms. Accessed at:
http://www.pall. com/main/medical/product.page?id=45154
14. Pall Corporation. March 2013. Declaration of Compliance: Pall-Aquasafe™ Disposable
Water Filter 31 Day Use - Tap Application. Accessed at:
http://www.pall.com/pdfs/Medical/A031FlR-Declaration-of-Compliance.pdf
15. Pall Corporation. Nov 2012. QPoint™ Tap Water Filter - USA. Accessed at:
http://www.pall.com/main/consumer-water/product.page71idHi8pwl57i
16. Pall Corporation. Feb 2009. Pall-Aquasafe™ Disposable Water Filter - Tap (AQ31F1S
and AQ31F1R) Field Evaluation Report.
17. Pall Corporation Aquasafe Medical Filters. See:
http://www.pall. com/main/medical/product.page?id=45154#
18. Pall Medical North American Sales Representatives Personal Communication. February
24, 2014.
19. Pfister ,S., Saner, D., Koehler, A. 2011. The environmental relevance of freshwater
consumption in global power production. International Journal of Life Cycle Assessment,
16(6): 580-591.
92
-------
20. U.S. Department of Energy: Buildings Technology Program. June 2012. Life Cycle
Assessment of Energy and Environmental Impacts of LED Lighting Impacts. Accessed
at: http://appsl.eere.energy.gov/buildings/publications/pdfs/ssl/2012_led_lca-pt2.pdf
21. U.S. Department of Energy: Buildings Technology Program. Oct 2013. Faucets.
Accessed at:
http://wwwl.eere.energy.gov/buildings/appliance_standards/product.aspx/productid/64
22. U.S. EPA. 2013. National Primary Drinking Water Regulations. See:
http ://water. epa. gov/drink/contaminants/index. cfm
23. U.S. EPA's Tool for the Reduction and Assessment of Chemical and Other
Environmental Impacts (TRACI). See: http://www.epa.gov/nrmrl/std/sab/traci/.
24. U.S. EPA. February 2013 .Quality Assurance Project Plan for Systems-Based
Sustainability and Emerging Risks Performance Assessment of Cincinnati Regional
Water Technology Innovations: Comparative Life Cycle Assessment and Cost Analysis of
Water Treatment Options. Prepared by Eastern Research Group, Inc. for U.S. EPA
Sustainable Technology Division, National Risk Management Research Laboratory.
93
-------
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
------- |