United States

Environmental Protection
Agency

Work Breakdown Structure-Based Cost Model

for Reverse Osmosis/Nanofiltration Drinking
Water Treatment

Office of Water (4607M)

EPA ***-*-*****
June 2019
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Table of Contents
1. Introduction ............................................................................................................................... 1
1.1 Background ...................................................................................................................... 1
1.2 Objectives ......................................................................................................................... 1
1.3 Organization of the Report ............................................................................................... 2
1.4 List of Abbreviations and Symbols in this Chapter ......................................................... 2
1.5 References ........................................................................................................................ 2
2. WBS Model Overview .............................................................................................................. 3
2.1 Model Structure ................................................................................................................ 3
2.2 The WBS Approach ......................................................................................................... 8
2.3 Model Use ...................................................................................................................... 10
2.3.1 Input Sheet Structure and Use................................................................................. 10
2.3.2 Common Inputs....................................................................................................... 13
2.3.3 Input Sheet Examples ............................................................................................. 16
2.3.4 Output Sheet Structure and Use .............................................................................. 18
2.3.5 Critical Design Assumptions Sheet Structure and Use ........................................... 18
2.3.6 Index Sheet Structure and Use ................................................................................ 21
2.4 General Cost Assumptions ............................................................................................. 22
2.4.1 Building Costs......................................................................................................... 24
2.4.2 Residuals Management Costs ................................................................................. 24
2.4.3 Indirect Capital Costs.............................................................................................. 25
2.4.4 Add-on Costs .......................................................................................................... 26
2.4.5 Annual O&M Costs ................................................................................................ 27
2.4.6 Total Annualized Cost ............................................................................................ 28
2.4.7 Updating and Adjusting Costs ................................................................................ 28
2.5 List of Abbreviations and Symbols in this Chapter ....................................................... 29
2.6 References ...................................................................................................................... 29
3. Reverse Osmosis and Nanofiltration Model ......................................................................... 31
3.1 Overview of the RO/NF Treatment Process .................................................................. 32
3.2 Input Sheet...................................................................................................................... 36
3.3 Model Assumptions Sheets ............................................................................................ 49
3.4 Influent and Booster Pumping Sheet .............................................................................. 51
3.5 Pre-treatment Sheet ........................................................................................................ 52
3.6 Feed Water Sheet ........................................................................................................... 53
3.7 Train Design Sheets ....................................................................................................... 53
3.7.1 Overview ................................................................................................................. 53
3.7.2 Train Design Sheet.................................................................................................. 54
3.7.3 Pressure Tuning Sheet............................................................................................. 55
3.7.4 Calc System and Calc One Element Sheets ............................................................ 56
3.8 Membrane Capital Sheet ................................................................................................ 59
3.9 Cleaning Sheet................................................................................................................ 60
3.10 Pipes and Structure Sheet ........................................................................................... 61
3.11 Instrumentation and Control Sheet ............................................................................. 62
3.12 Residuals Management Sheet ..................................................................................... 62
3.13 O&M and HVAC Sheets ............................................................................................ 63
3.14 Indirect Sheet .............................................................................................................. 64

i
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

3.15 Output Sheet ............................................................................................................... 64

3.16 Ancillary Model Components .................................................................................... 65
3.17 List of Abbreviations and Symbols in this Chapter.................................................... 65
3.18 References .................................................................................................................. 66
Appendix A. Valves, Instrumentation and System Controls .................................................. 68
Appendix B. Building Construction Costs ................................................................................ 74
Appendix C. Residuals Management Costs .............................................................................. 83
Appendix D. Indirect Capital Costs .......................................................................................... 92
Appendix E. General Assumptions for Operating and Maintenance Costs ........................ 115

ii
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

1. Introduction
This report is one of a series of reports describing cost models for drinking water treatment
technologies. Most of these technologies are used in drinking water systems to remove or destroy
pollutants such as arsenic, radon, disinfection byproducts, sulfates, hardness and waterborne
pathogens. In addition, several of these technologies can be used as add-on technologies to
existing treatment systems. For example, some of the technologies can be installed to provide
pre-oxidation to improve contaminant removal efficiency by subsequent treatment processes.

1.1 Background
The Safe Drinking Water Act Amendments of 1996, as well as a number of other statutes and
executive orders, require that the U.S. Environmental Protection Agency (EPA or the Agency)
estimate regulatory compliance cost as part of its rulemaking process. EPA developed the models
described in this document to assist in fulfilling this requirement. In other words, the primary
purpose of these models is to aid EPA in estimating national compliance costs. The models
might be acceptable, however, for other uses (e.g., developing a preliminary site-specific
estimate for a water system) if sufficient care is taken to account for site- or project-specific
factors appropriate to the intended use.

The compliance cost models described in this document differ from the drinking water cost
models previously used by the Agency in that the new models are based on a work breakdown
structure (WBS) approach to developing cost estimates. In general, the WBS approach involves
breaking a process down into discrete components for the purpose of estimating unit costs. EPA
pursued this approach as part of an effort to address recommendations made by the Technology
Design Panel, which convened in 1997 to review the Agency’s methods for estimating drinking
water compliance costs (U.S. EPA, 1997). 1

1.2 Objectives
In developing WBS-based models for estimating drinking water treatment system costs, EPA
had the following objectives:

• Transparency of process design and cost

• Defensibility of design criteria and assumptions
• Ease of use and updating
• Modularity of components for use with centralized cost database.

The Agency determined that the best way to meet these goals was to develop spreadsheet-based
engineering models drawing from a central database of component unit costs. Each engineering
model contains the work breakdown for a particular treatment process and preprogrammed
engineering criteria and equations that estimate equipment requirements for user-specified design
requirements (e.g., system size and influent water quality). Each model also provides unit and
total cost information by component (e.g., individual items of capital equipment) and totals the

1
The panel consisted of nationally recognized drinking water experts from U.S. EPA, water treatment consulting
companies, public and private water utilities, suppliers, equipment vendors and Federal and state regulators in
addition to cost estimating professionals.

1
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

individual component costs to obtain a direct capital cost. Additionally, the models estimate add-
on costs (permits, pilot study and land acquisition costs for each technology), indirect capital
costs and annual operating and maintenance (O&M) costs, thereby producing a complete
compliance cost estimate.

1.3 Organization of the Report

This report is organized as follows:

• Chapter 2 provides an overview of the general model components and the methods used
in these components to estimate treatment system costs.
• Subsequent chapters describe the individual models, design criteria and assumptions for
the selected treatment technologies.
• Appendices provide additional information on methods EPA used to estimate design
requirements and costs for specific components, such as buildings, system controls,
indirect capital costs and annual O&M costs.

1.4 List of Abbreviations and Symbols in this Chapter

EPA U.S. Environmental Protection Agency
O&M operating and maintenance
WBS work breakdown structure

1.5 References
U.S. Environmental Protection Agency (U.S. EPA). 1997. Discussion Summary: EPA
Technology Design Workshop. Washington, D.C.: U.S. EPA, Office of Groundwater and
Drinking Water.

2
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

2. WBS Model Overview

This chapter includes the following sections:

• An overview of how the models are structured (Section 2.1)

• A description of how this structure was developed using the work breakdown structure
(WBS) approach (Section 2.2)
• A brief users guide describing how to operate the models (Section 2.3), including
documentation of general design assumptions
• Documentation of the general cost assumptions incorporated in all of the models (Section
2.4).

2.1 Model Structure

The WBS-based engineering models integrate the following structural features to generate
treatment cost estimates:

• Treatment component selection, design and cost output based on a WBS approach
• Process design based on state-of-the-art techniques and generally recommended
engineering practices (GREPs)
• A centralized reference database containing unit costs for components and reference
tables for component sizing and chemical properties.

Exhibit 2-1 shows how these features are integrated in a series of spreadsheets that include an
Excel workbook for each technology and a central cost and engineering reference database (the
WBS cost database). 2 An input sheet allows the user to define treatment requirements such as
system design and average flows, target contaminant and raw water quality. Exhibit 2-2
provides an example of an input spreadsheet. The information provided via the input sheet
interacts with three critical design assumptions sheets (one each for process design, operating
and maintenance [O&M] and indirect capital costs) to generate inputs to the engineering design
sheets. Although the critical design assumption values are based on GREPs and can be used
without modification, the user can also revise these values to reflect site-specific requirements.
Each model also has a predetermined list of treatment equipment needs (e.g., tanks, vessels and
instrumentation) identified using the WBS approach. The engineering design sheets calculate
equipment quantity and size requirements based on the treatment needs and critical design
assumptions. The technology chapters of this report describe technology-specific content and
function of each sheet. General design and cost assumptions are described in Sections 2.3.5 and
2.4.

2
EPA maintains the central WBS cost database in a separate Excel workbook. The WBS cost database itself is not
provided along with the publicly released WBS models. Instead, for ease of review and to maintain vendor
confidentiality, relevant cost and engineering data have been extracted from the database and included directly in the
WBS model workbooks. Thus, users can review (and adjust, if needed) the information from the central cost
database in the same manner as other WBS model inputs and assumptions.

3
WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 2-1. Structure of the WBS Models

4
WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 2-2. Sample of Input Spreadsheet

5
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 2-3 shows an example of an output spreadsheet. The output sheet summarizes the results
of the calculations performed by the engineering design sheets, listing size and quantity required
for each item of equipment and the corresponding unit cost from the database. The output sheet
multiplies unit cost by quantity to determine total component cost for each WBS component. The
output sheet also lists the estimated useful life of every WBS component. The models use the
component useful lives in estimating total annualized cost (see Section 2.4.6).

For many of the components, there are optional materials, all of which are illustrated on the
output worksheet. For example, pressure vessels can be constructed with different types of body
material (stainless steel or carbon steel) and different types of internal materials (stainless steel or
plastic). Where there are optional materials, the output sheet selects from among these materials.
The specific selections are determined by input values and documented in the “use?” column of
the output worksheet. Direct capital cost is the sum of the selected component costs.

The output sheet also contains sections that calculate add-on costs, indirect capital costs, annual
O&M costs and total annualized cost. Annual O&M costs are based on the annual requirements
calculated on the O&M sheet. Indirect capital costs for certain items (standby power,
geotechnical, site work and yard piping) are based on calculations performed by the indirect
sheet. Other indirect capital costs and add-on costs are based on assumptions described in
Sections 2.4.3 and 2.4.4. Section 2.4.6 describes the calculation of total annualized cost.

The output sheet obtains unit costs (both capital and O&M) either from the central WBS cost
database or from estimated equipment cost curves. All of the treatment technology models use
information from the WBS cost database, which consists of a series of lookup tables that contain
costs by equipment or O&M element type and size. The database also provides useful life
estimates and documents the source of information. The central WBS cost database also contains
several tables that are used by the engineering design sheets of each model. For example, these
tables include information used in selecting pipe diameters, footprint for pumps and chemical
properties.

The WBS cost database itself is not provided along with the publicly released WBS models.
Instead, for ease of review and to maintain vendor confidentiality, relevant cost and engineering
data have been extracted from the database and included directly in the WBS model workbooks.
Thus, users can review (and adjust, if needed) the information from the central cost database in
the same manner as other WBS model inputs and assumptions.

6
WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 2-3. Sample of Output Spreadsheet

7
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

2.2 The WBS Approach

These models represent improvements over past cost estimating methods by increasing
comprehensiveness, flexibility and transparency. By adopting a WBS-based approach to identify
the components that should be included in a cost analysis, the models produce a more
comprehensive assessment of the capital requirements for a treatment system. The models are
flexible in that users can change certain design parameters; warning messages indicate when user
inputs violate GREPs or logical functions. The transparent structure of each model allows users
to see how costs are built up from component unit costs to total treatment costs, which enables
users to identify cost drivers and determine whether the input assumptions generate a cost-
effective treatment design. Users also can perform sensitivity analyses showing how changes in
water quality parameters, chemical feed doses and equipment configuration affect cost.
Unlike prior EPA models, which used a variety of cost build-up methods, the WBS-based
engineering models have been developed using a consistent framework. Exhibit 2-4 shows this
framework. For each technology, the result is an engineering spreadsheet model that combines
user-identified inputs with pre-programmed engineering criteria and equations to generate
appropriate treatment design and equipment requirements. The models also result in a system-
level cost estimate for regulatory cost analysis.
Exhibit 2-4. Framework for Developing the WBS-Based Models
Step Process
Step 1 Identify the treatment requirements based on the contaminant requiring removal, the flow for which treatment is
required, the influent water quality and treated water quality requirement, and then select a treatment technology
or combination of technologies capable of meeting the requirements.
Step 2 Develop the general design assumptions that apply to all the technologies (e.g., chemical storage capacity).
Step 3 Develop site- and technology-specific design assumptions that might affect treatment performance and, thereby,
design requirements (e.g., assumptions related to influent water constituents such as alkalinity or water quality
parameters such as pH).
Step 4 Construct a typical process flow diagram or P&ID showing the main unit processes for the technology and identify
equipment requirements.
Step 5 Calculate the equipment requirements, including dimensions and quantities, for the core elements of each unit
process. At each component (or group) level, identify choices of material (e.g., stainless steel or PVC pipe
material).
Step 6 Link the treatment equipment requirements to a database that contains unit costs by equipment type, size and
material. Multiplying the unit costs by the dimension and quantity requirements developed in Step 5 provides the
component-level design costs.
Step 7 Tally the costs of the selected components to determine direct capital cost.
Step 8 Develop and add indirect and add-on costs to determine total system capital cost.
Step 9 Develop operation and maintenance cost estimates.

The WBS approach provides EPA with a consistent method for identifying components to
include in a cost estimate. For each technology, the WBS approach develops a process and
instrumentation diagram (P&ID) or a typical schematic layout showing the main unit processes
needed to achieve the contaminant removal goals.
Exhibit 2-5 provides examples of several classes of components that can be included in a P&ID.
The models often include further breakdown for alternative materials of construction for each
component, because costs can differ substantially across materials. For example, most pipes can

8
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

be constructed of stainless steel, steel, polyvinyl chloride (PVC) or chlorinated PVC. Stainless
steel piping can cost twice as much as PVC.
Exhibit 2-5. Component Classes Included in the WBS Inventory
Component Classes Example Components
Vessels Pressure vessels
Tanks/basins Storage
Backwash
Mixing
Contact
Flocculation
Sedimentation
Filtration
Pipes Process
Backwash
Chemical
Inlet/outlet
Bypass
Valves (see Appendix A for further details) Check (one-way)
Motor- or air-operated
Manual
Pumps Booster
Backwash
High-pressure (for membrane systems)
Chemical metering
Mixers Rapid
Flocculation
Inline static
Instrumentation (see Appendix A for further details) Pressure gauge
Level switch/alarm
Chlorine residual analyzer
Flow meter
pH meter
Air monitor/alarm
High/low pressure alarm
Gas flow meters—rotameters
Scales
System controls (see Appendix A for further details) Programmable logic control units
Operator interface equipment
Controls software
Chemicals Acids
Bases
Coagulants and coagulant aids
Antiscalants
Corrosion control
Oxidants and disinfectants
Treatment media Activated alumina
Activated carbon
Membranes
Sand
Resins

9
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Component Classes Example Components

Building (see Appendix B for further details) Structure
Heating and air conditioning systems
Concrete pad
Indirect Capital Components (see Appendix D for further Geotechnical investigations
details) Standby power generators

The level of component detail (and by implication, design detail) in Exhibit 2-5 indicates that
the WBS-based approach is more sophisticated, and potentially more time consuming, than the
factored or parametric cost estimating methods used in earlier efforts. Nevertheless, the
Technology Design Panel considered it the right approach to developing unit costs for policy
analysis. Furthermore, EPA believes that developing unit cost models that are more
comprehensive, flexible and transparent will facilitate the policy analysis process by addressing a
frequent topic of dispute over regulatory cost estimates. Finally, the WBS-based models are
driven by technical scope and selection of suitable equipment and material to achieve a defined
treatment objective. This approach is superior to cost estimating methods that are not defined by
a desired treatment level or that cannot be changed easily to reflect raw water quality.

2.3 Model Use

This section provides basic guidance on operating the WBS technology models. As discussed
above, each model is an Excel workbook comprising a series of spreadsheets. In general, users
need only be concerned with the input sheet and output sheet, although advanced users might
also wish to examine the critical design assumptions spreadsheets.

2.3.1 Input Sheet Structure and Use

The input sheet in each of the technology models is similar to that pictured in Exhibit 2-2. A
step-by-step input process allows the user to quickly generate costs for standard designs built
into the model, modify those designs or construct an alternative design.

Overview of the Input Process

Many models require basic information
from the user before choosing an
appropriate standard design. For example,
contaminant selection is the first choice
that must be made in several of the
models. Such choices are made using a
drop-down list at the top of the input
sheet.

After making any basic, top-level choices,

the user can click on one of the eight
standard design buttons. Each button
corresponds to a system size category in
the flow characterization paradigm
described below in Exhibit 2-6. The
model will populate all inputs with values

10
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

appropriate for the selected design, then compute all costs. The direct capital cost, total capital
cost and annual O&M cost are displayed on the input sheet; details are available on the output
sheet (see Section 2.3.4). More information on the standard designs is provided below.

Exhibit 2-6. Standard Flow Rate Categories Used in WBS Standard Designs
Size Category Population Served Design Flow (MGD) Average Flow (MGD)
1 25 to 100 0.030 0.007
2 101 to 500 0.124 0.035
3 501 to 1,000 0.305 0.094
4 1,001 to 3,300 0.740 0.251
5 3,301 to 10,000 2.152 0.819
6 10,001 to 50,000 7.365 3.200
7 50,001 to 100,000 22.614 11.087
8 Greater than 100,000 75.072 37.536

The standard designs, with their corresponding buttons, are primarily for EPA’s use in estimating
costs for a median sized system in each size category, although some users may find them useful
as a starting point (see the examples in Section 2.3.3). The user can modify the standard designs
after clicking one of the buttons by entering values in the gold and blue input cells, under the
“Manual Inputs” heading on the input sheet. Alternately, many users will want to click the button
marked “CLEAR FOR MANUAL ENTRY” and enter all of the input values by hand. In any
case, the manual inputs section contains several types of cells:

• Required user inputs, highlighted in gold

• Optional user inputs, highlighted in blue
• Greyed-out inputs, which are not required for a given design
• Information and guidance, with text in green.

Some inputs, such as system flows, must contain a numeric value. Others have a drop-down
arrow that appears when the cursor is positioned in the input cell. These cells must contain one of
the drop-down values. Required inputs must be populated; optional inputs can be left blank to
accept model defaults or changed by the user to examine the effect of different assumptions. The
Autosize button, described below, is available in some models to facilitate design.

The input sheet in each model verifies user inputs against certain design constraints that reflect
GREPs. If user inputs result in designs that violate these constraints, a warning message appears
on the input sheet, explaining which input value needs to be corrected. In addition, the message
“Input Incomplete—Check for Error Messages Below” appears at the top of the input sheet.

Once all inputs are complete and the

model has verified that they meet design
constraints, the message at the top of the
input sheet changes to “Input
Complete—Press ‘Generate Results’.”
The user must click the “Generate
Results” button to tell the model to
generate costs. Once the user has clicked

11
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

the button, the message at the top of the model changes to “Input Complete—Results Ready,”
and total costs are displayed on the input sheet. The output sheet provides more details for the
total costs.

Standard Designs
The input sheet in each of the technology
models contains up to eight buttons,
which correspond to the eight standard
flow sizes in the flow characterization
paradigm for public water systems (see
Exhibit 2-6). These buttons populate all
of the input fields with appropriate values
for the selected design flow. The values in
each standard design meet all relevant
design constraints. Each model includes a
separate sheet, entitled “standard inputs,”
that documents the specific input values
included in each standard design.
Advanced users can adjust the standard
designs by changing the values on the
standard input sheet. For example, a user
could change all the standard designs to use high cost components, rather than the default of low
cost components (see Section 2.3.2 under “Component Level”), by changing values in the
appropriate column on the standard input sheet. The standard input sheet highlights values that
have been changed by the user and includes a button (“Reset to Defaults”) that resets the
standard designs back to their original settings. Users that make significant adjustments to the
standard designs should take care to verify that their new designs still meet design constraints by
checking for warning messages on the input sheet after each new design is run.

The Autosize Routine

The models also can be used to estimate
costs for systems with design flows other
than the eight standard sizes. To aid in
developing designs for other flows, some
models include a button labeled
“Autosize.” This button activates a
computer-aided design routine that
attempts to find a design meeting all
relevant design constraints for a given design and average flow. For example, the user could
change design flow to 3 million gallons per day (MGD) and average flow to 1 MGD, then click
the autosize button. This would populate some input fields with values that are both appropriate
for a 3 MGD system and that meet all design constraints. More information on the autosize
routines, including details on which inputs are and are not populated, is available in the
technology-specific chapters of this document.

12
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

In the rare case that the autosize routine cannot find a design meeting all constraints, it will
display a pop-up warning message. This does not mean that it is impossible to design a system
for the selected size. The user might still be able to develop a design by manually adjusting the
input values, paying careful attention to the warning messages on the input sheet. It might be
necessary to relax some of the design constraints by adjusting values on the critical design
assumptions sheet.

Manual Input and “Generate Results”

All of the models allow the user to enter input values by typing them directly into the appropriate
fields on the input sheet. Users can develop complete designs from scratch, populating all the
input fields manually. Users also can adjust designs generated by the standard design or autosize
buttons, by adjusting one or more input fields manually after clicking one of these buttons. In
either case, after completing the manual changes, users should do two things:

• Verify that no warning messages appear to ensure that the design meets all relevant
constraints
• Click the button labeled “Generate Results.”

The second step is necessary to tell the

models that the design process is
complete and to select the appropriate
items of equipment for inclusion in total
costs on the output sheet. This step is
particularly important if the system
automation or component level inputs
are adjusted manually, because these
inputs have a significant impact on the selection of equipment. To ensure correct calculation of
costs, however, users should click the “Generate Results” button after completing manual
changes to any of the inputs. It is not necessary to click this button when the input sheet message
reads “Input Complete—Results Ready.” This message will appear, for example, when the
standard designs or autosize routine are used without subsequent manual changes to input values.
The standard design buttons and the autosize button automatically incorporate the “Generate
Results” step, telling the models to select the appropriate items of equipment.

2.3.2 Common Inputs

The user inputs in each model are largely technology-specific and are described in detail in the
technology chapters of this document. There are certain inputs, however, that are common to all
of the technology models. These common inputs are described below.

13
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Design and Average Flow

Each model needs the design and average flow to determine the size and number of treatment
components needed. Design flow is the peak instantaneous flow of product water from a
treatment system, while average flow is the annual average flow, taking into account daily and
seasonal variations in demand.

Design flow can be entered in MGD or in gallons per minute (gpm). In either case, the design
flow is meant to represent a maximum instantaneous flow. Average flow can be entered in
MGD, in gpm or as a percentage of design flow.

The standard design functions included in each model (see above) can populate design and
average flow with values based on the flow characterization paradigm for public water systems.
The flow paradigm includes eight model size categories, as shown in Exhibit 2-6. These size
categories represent populations ranging from 25 persons to greater than 100,000 persons. Based
on the values in Exhibit 2-6, the ratio of average flow to design flow ranges from 25 percent for
very small systems to 50 percent for large systems.

Component Level

Each model includes an optional input that determines whether the cost estimate generated is a
low, medium or high cost estimate. This input, labeled “component level” or “cost level,” drives
the selection of materials for items of equipment that can be constructed of different materials.
For example, a low cost system might include fiberglass pressure vessels and PVC piping. A
high cost system might include stainless steel pressure vessels and stainless steel piping. The
component level input also drives other model assumptions that can affect the total cost of the
system, including assumptions about system automation (see “System Automation” below),
building quality and heating and cooling (see Appendix B). 3 If the component level input is left
blank, the models will generate a low cost estimate. The user can change this input to select a
medium or high cost estimate.

3
In some cases (e.g., the membrane models, which are under development), this input also determines the source
water quality that the model treats. In these models the input is called the “cost level.”

14
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

System Automation

As described in Appendix A, control of drinking water treatment systems can be manual,

automated or semi-automated. The method of control can have a significant impact on both
capital and O&M costs. Each model includes an optional input that allows the user to select from
among the three control options. If the system automation input is left blank, the control option
selected is determined by the system size and the component level input selected (see above),
using the logic shown in Exhibit 2-7. The user can change the system control input to force the
design of a system with manual, automated or semi-automated control.

Exhibit 2-7. Default Assumptions for System Control

System Size (Design Flow) System Size (Design Flow)
Component Cost Level Selected Less than 1 MGD 1 MGD or greater
Low Manual Manual
Medium Manual* Automated
High Automated Automated
* Automated for some models.

Include Buildings?

By default, the WBS models include the capital cost of buildings to house the treatment system,
as discussed in Section 2.4.1 and Appendix B. Each model includes an optional input that allows
the user to exclude the capital cost of buildings. If the user excludes the capital cost of buildings,
the model also excludes the O&M cost of building maintenance and lighting.

Include Heating, Ventilating and Air Conditioning (HVAC)?

By default, the WBS models choose whether to include the cost of heating and cooling systems
depending on system size, building structure type and user input for component level, as
discussed in Section 2.4.1 and Appendix B. Each model includes an optional input that allows
the user to override the model’s default selection and choose to include or exclude the cost (both
capital and O&M) of HVAC systems.

Include Land?
Regardless of whether a system needs to purchase additional land on which to build the new
treatment train, there is an opportunity cost associated with using land for water treatment rather

15
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

than an alternative use. By default, the WBS models include an add-on cost for land, as
discussed in Section 2.4.4. Each model includes an optional input that allows the user to exclude
the add-on cost for land.

2.3.3 Input Sheet Examples

Several examples are presented here to clarify the use of the WBS model input sheet. The
examples refer to particular technology models. Detailed information about the inputs for these
models can be found in the appropriate technology-specific chapters.

Standard Design
The simplest way to generate a design is by use of the standard design buttons. Suppose that a
user wishes to estimate costs for a system designed to treat trichloroethylene (TCE) using
granular activated carbon (GAC), serving a population of approximately 8,000 people. The
following are step-by-step instructions for using the adsorptive media model to generate such a
cost estimate:

1. Open the Excel workbook named “WBS GAC.xlsm.” 4 Depending on your settings and
version of Excel, a message might appear regarding “active content” in the workbook.
For the models to function properly, macros must be enabled. Take the appropriate steps
to enable macros (for example, clicking “Options” and selecting “Enable this content,”
depending on your version of Excel).

2. Navigate to the input sheet by clicking on the tab labeled “INPUT” at the bottom of the
Excel window. (It is also possible to page through the sheets by pressing Ctrl-Page Up
and Ctrl-Page Down.) Scroll to the top of the input sheet.

3. The GAC model requires that the user first choose the contaminant. Select “TCE” from
the “Select Contaminant” dropdown list.

4. The GAC model also requires that the user choose between pressure and gravity designs
(see the appropriate technology chapter for discussion of the difference between design
types). Select “Pressure” from the “Select Design Type” dropdown list.

5. The user wishes to use a standard design appropriate for a population of 8,000 people.
Exhibit 2-6 indicates that size category 5, with a design flow of 2.152 MGD, is
appropriate for such a system. Therefore, click on the design button labeled “2.152 MGD
standard design.” After a few seconds, the model will display the message “Using this
design” next to the design button and “Input Complete—Results Ready” underneath the
buttons. It displays the direct capital cost, total capital cost and annual O&M cost on the
input sheet.

6. If desired, scroll down on the input sheet to see what inputs are used for the standard
design. For instance, the 2.152 MGD standard design for GAC treating TCE with a
pressure design uses a design flow of 2.152 MGD and an average flow of 0.819 MGD. It

4
Note that your model file name might vary. It likely will include a date following the model title (e.g., “WBS GAC
042514.xlsm” for April 25, 2014).

16
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

assumes a carbon life of 66,600 bed volumes and a total theoretical empty bed contact
time (EBCT) of 7.5 minutes.

Modified Standard Design

Suppose that the user wishes to design a GAC system treating TCE for a population of 1,000,
using source water that entails a different carbon life and EBCT than that assumed in the
standard designs (e.g., because the source water contains a higher initial concentration of TCE).
The user determines that the source water characteristics entail a carbon life of 40,000 bed
volumes and an EBCT of 10 minutes. The following are step-by-step instructions for using the
GAC model to generate such a cost estimate:

1. Open the Excel workbook named “WBS GAC.xlsm” 5 and take the appropriate steps to
enable macros (see Step 1 described in the “Standard Design” section above). Navigate to
the input sheet, scroll to the top of that sheet and select “TCE” and “Pressure” from the
appropriate dropdowns (see Steps 2, 3 and 4 described in the “Standard Design” section
above).

2. The user wishes to design a system for a population of 1,000 people. Exhibit 2-6
indicates that size category 3, with design flow 0.305 MGD, is appropriate for this
population, so start by clicking the “0.305 MGD standard design” button.

3. The user wishes to design a system with a carbon life of 40,000 bed volumes. Scrolling
down the input sheet, note that the standard design uses an input carbon life of 66,600
bed volumes. Type the number 40,000 in the gold input cell to change the carbon. Note
that the green informational text below the input cell changes to show the number of
months between regenerations. Note also that the message above the manual inputs
changes to “Input Complete—Press ‘Generate Results’” to indicate that costs have not
been updated for your new input.

4. The user wishes to design a system with an EBCT of 10 minutes. Scroll down to the cell
labeled “Theoretical Empty Bed Contact Time” and enter the number 10.

5. Changing the EBCT will change the optimal vessel geometry. To quickly estimate costs
for this new EBCT, click the “Autosize” button next to the inputs for vessel geometry.
The input values will flicker briefly while the model tries several different values and
then settles on a new value. Because the Autosize button was clicked, it is not necessary
to click the “Generate Results” button; the message above the manual inputs reads “Input
Complete—Results Ready,” and the total costs are displayed on the input sheet.

Suppose that the user also wishes to estimate a high-end cost for this system. In this case, take
the following additional steps:

5
Again, your model file name might vary. It likely will include a date following the model title (e.g., “WBS GAC
042514.xlsm” for April 25, 2014).

17
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

6. Scroll down and place the cursor in the input cell labeled “Component Level.” A
dropdown arrow appears to the right of the cell. Click on the arrow and choose “high
cost.”

7. Scroll back to the top of the sheet. Note that the sheet indicates that the user must click
“Generate Results.” Click that button. The model displays costs for the high-end system.
To see what components are included, switch to the Output sheet and examine the details.

2.3.4 Output Sheet Structure and Use

The output sheet in each of the technology models is similar to that pictured in Exhibit 2-3. In
addition to the details described in Section 2.1, the output sheet includes several important totals:

• Process cost, which is the sum of the installed capital cost of all equipment required for
the treatment process
• Building cost, which is the sum of the installed capital cost of all buildings and the
concrete pad
• Direct capital cost, which is the sum of the process and building costs
• Total capital cost, which is the total of the direct capital cost, the indirect capital costs and
add-on costs (see Sections 2.4.3 and 2.4.4)
• Annual O&M cost (see Section 2.4.5)
• Total annualized cost (see Section 2.4.6).

The capital equipment section of the output sheet includes a column labeled “Use?” This column
tells the model which line items to include in the direct capital cost. Specifically, items with a
value of 1 in the “Use?” column are included in the total; items with a value of 0 or a blank are
not included in the total. Advanced users can manually adjust this column to include or exclude
certain items of equipment. For example, a user could examine process costs without booster
pumps by changing the “Use?” value to 0 for those pumps. The “Generate Results” button,
which is present on both the input and output sheets, will reset the “Use?” values back to pre-
programmed default values, as driven by system size and input values.

The output sheet also includes a button labeled “Record Output in a New Workbook.” This
button generates a complete copy of the output sheet that will not change. Using this button
allows users to record the detailed design output for comparison purposes. For example, a user
could record the output from the standard design for 0.03 MGD, then select the 0.124 MGD
standard design and compare the output results for the two designs.

2.3.5 Critical Design Assumptions Sheet Structure and Use

Each of the technology models includes at least three critical design assumptions sheets:

• One for process and building design assumptions

• One for assumptions used in calculating annual O&M costs
• One for assumptions used in calculating certain indirect capital costs

Some models include additional critical design assumptions sheets (e.g., in the aeration models,
for assumptions associated with off-gas treatment).

18
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

These sheets contain design constraints and structural and chemical engineering assumptions
based on GREPs. Users can review these sheets for details on significant assumptions used in the
models. Advanced users might want to modify certain assumptions, particularly if adapting a
model for use with a source water quality different than assumed in the standard designs or to
reflect site-specific conditions. Most of the assumptions include a comment column explaining
the use of the assumption and/or providing guidance on appropriate values.

Most of the significant design assumptions are technology-specific and discussed in detail in the
technology chapters of this report. However, there are certain assumptions that are common to
many of the models. Exhibit 2-8 summarizes the general design assumptions that are common
across most of the models. As Exhibit 2-8 indicates, these assumptions are based on a
combination of sources, including standard design handbooks, engineering textbooks and
comments of external reviewers. Note that some of the general design assumptions (and some
technology-specific assumptions, as discussed in the relevant technology chapters) differ for
small versus large systems. In general, these differences are because small systems can often be
built as packaged, pre-engineered or skid-mounted systems. In most cases, the different design
and cost assumptions for small systems are based on comparison of model outputs with as-built
designs and costs for actual small treatment systems.

The user can change some of the assumptions shown in Exhibit 2-8 by editing the critical design
assumptions sheet; others can be edited in the data extracted from WBS cost database. The final
column of Exhibit 2-8 provides guidance on how to change each assumption. For example, the
design of pumps for any treatment system is based on the peak flow requirements of the system,
including a safety factor. As specified in Exhibit 2-8, the critical design assumptions sheet
assumes a safety factor of 25 percent. A user could change this factor based on an actual pump
performance curve.

Exhibit 2-8. General Design Assumptions Used in the WBS-based Models

Element Assumption Can be changed by:
Influent pumps Include flooded suction Replacing unit costs or cost coefficients
extracted from the WBS cost database
All pumps Design flow incorporates a safety factor of Editing the critical design assumptions
25 percent sheet of each technology
Access space for pumps Provide a minimum of 4 feet of service Editing the critical design assumptions
space around three sides of each unit, sheet of each technology
assuming the fourth side can share access
space with relevant tanks or vessels
Pipe size Based on a maximum of 3 feet of head Editing the engineering lookup table
loss per 100 feet of pipe extracted from the WBS cost database
Process pipe size Based on maximum flow to each unit (not Cannot be changed
total system flow)
Tank and pressure vessel Based on design capacity, freeboard and Cannot be changed
capacity standard manufactured sizes
Pressure vessel diameter Based on user input, within limits specified Changing user inputs (for diameter) and
on a technology-specific basis editing the critical design assumptions
sheet of each technology (for constraints)
Storage tank diameter Assumes a cylindrical design, with Cannot be changed
diameter equal to one half of the height

19
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Element Assumption Can be changed by:

Access space for tanks and Provide service space around each unit Editing the critical design assumptions
pressure vessels equal to its diameter (half its diameter forsheet of each technology (only maximum
small systems), to a maximum of 6 feet can be changed)
Process vessels and basins, all Multiple units required to protect from Editing the critical design assumptions or
pumps and chemical feed single point failure input sheet of each technology (depending
systems on the specific item)
Chemical storage Storage requirement based on 30-day Editing the critical design assumptions
delivery frequency sheet of each technology
Concrete pad under heavy 1 foot thick for large systems, 6 inches Editing the critical design assumptions
equipment thick for small systems sheet of each technology
Office space 100 square feet per employee for large Editing the critical design assumptions
systems (excluded for small systems) sheet of each technology
Sources: U.S. EPA (1997); AWWA (1990); AWWA/ASCE (1998); Viessman and Hammer (1993); GREPs; and information from
manufacturers and technology experts who reviewed model critical design assumptions.

Cost Estimation Method

Equipment unit costs can be derived in one of two ways. The first (and recommended) method
uses component-specific cost equations developed from unit costs collected from equipment
vendors. The component cost equations are best-fit equations (developed using statistical
regression analysis across the sizes available for each item) that estimate the unit cost of an item
of equipment as a function of its size. Under the cost equation option, the models will generate
unit costs for each item of equipment by applying the appropriate cost equation to the exact size
determined by the design calculations.

The second method uses unit cost lookup tables extracted from the WBS cost database. These
lookup tables are based on quotes from equipment manufacturers for discrete equipment sizes.
To maintain vendor confidentiality, the tables do not identify the individual vendors associated
with the quotes and the unit costs typically are averages across multiple vendors. Under the
lookup table option, for each item of equipment, the models will search the appropriate lookup
table to locate a unit cost that best meets the design requirements for the component. In general,
this means that the models will select the discrete equipment size for each item of equipment that
is equal to or greater than the size determined by design calculations.

Each model includes a critical design assumption, labeled “cost estimating method,” that
determines the method used to derive equipment costs. By default, the assumption is set to 1, to
use the component-specific cost equations. The user can set the assumption to a blank value to
select the lookup table method. EPA believes the cost equations method is most appropriate for
generating national cost estimates and for most user-specified designs. Using the equations,
instead of the price quotes, allows the models to generate unit costs for equipment of the exact
size determined by the design calculations. For example, a WBS model design might require a
250 gallon steel tank, but the available price quotes might be limited to 100 gallon, 500 gallon
and various larger sizes. The cost equation for steel tanks will allow the WBS model to generate
a unit cost for the intermediate sized 250 gallon tank. The lookup table method would use the
cost for the 500 gallon tank. The models retain the lookup table method for users who wish to
examine the specific cost data points on which the component-specific cost equations are based.

20
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

2.3.6 Index Sheet Structure and Use

Each technology model includes an index of all inputs and critical design assumptions, including
hyperlinks to their locations. Exhibit 2-9 shows an example of the index sheet. The sheet
provides an alphabetized list of all inputs and assumptions. Due to the great number of inputs
and assumptions in the WBS models, the Find feature in Excel can be useful in locating a
specific input or assumption.

Next to the description of each input or assumption is a blue, underlined hyperlink. It shows the
internal name of the input or assumption used in the engineering formulas throughout the WBS
model. Clicking on the hyperlink takes the user to the cell where the assumption can be viewed
or adjusted.

21
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 2-9. Sample of Index Spreadsheet

2.4 General Cost Assumptions

An important feature of the WBS models is that they build up cost estimates from component-
level data. Each model shows the user the cost build-up, which makes the cost estimates more
transparent, giving the user an opportunity to evaluate the impact of design and unit cost
assumptions on treatment costs. There are several types of costs that need to be aggregated into a
total cost estimate: equipment costs, building cost, residuals discharge cost, indirect capital costs,
add-on costs and annual O&M costs. The sections below describe how each type of cost enters
the WBS models.

22
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The build-up process for equipment costs is straightforward. The design sheets in the model
generate the required dimensions and quantities for each item in the WBS list of equipment
components and materials. Then, the model obtains unit costs to match the component size and
material (e.g., a 10-inch diameter PVC pipe or a 4,000-gallon steel backwash tank). The model
multiplies unit costs by the quantity estimate (e.g., 30 feet of pipe or 2 tanks) to obtain total
component costs. Direct capital cost equals the sum of these costs across the selected
components, including costs for treatment equipment and buildings.

The models enable equipment unit costs to be derived in one of two ways (using lookup tables or
cost equations, as described in Section 2.3.5 under “Cost Estimation Method”). Regardless of the
method used, the estimates are intended to provide enough information to establish a budgetary
or preliminary cost estimate. Therefore, although the model results are point estimates shown to
the nearest $1, this precision is not meant to imply that the results are accurate to $1. Instead,
EPA’s goal is for the resulting costs to be within +30 percent to -15 percent of actual cost. To
validate the engineering design methods used by the models and assess the accuracy of the
resulting cost estimates with this goal, EPA has subjected the individual models to a process of
external peer review by nationally recognized technology experts. The technology-specific
chapters of this document include a discussion of peer reviewer opinions on the accuracy of each
model’s results. Users are encouraged to review all documentation, modify inputs and
assumptions as appropriate to their specific purpose, and form their own informed opinions about
the accuracy and suitability of the results.

Consistent with providing a budgetary or preliminary cost estimate, WBS models contain several
cost-related assumptions that allow the models to produce costs for some components without
having detailed site-specific information (e.g., pipe fitting sizes). Exhibit 2-10 summarizes these
assumptions.

23
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 2-10. General Equipment Cost Assumptions

1. Costs are preliminary estimates based on major components as shown on piping and instrumentation diagrams or typical
layout drawings. Costs include consideration of package plants where relevant (see model-specific chapters for more
details).
2. All equipment costs include costs of transportation and installation.
3. All equipment costs are based on cost quotes from manufacturers or RSMeans database.
4. Long-term storage of chemicals (greater than 30 days) is not taken into account unless specifically mentioned.
5. Cost of waste disposal (residuals) is accounted for using the methods outlined in Section 2.4.2 and Appendix C.
6. Building layout is for the process itself, with room for operation, maintenance and replacing equipment, if needed.
7. Building costs are estimated using unit costs per square foot (see Section 2.4.1 and Appendix B for more details).
8. Costs for a reinforced-concrete pad floor to handle equipment loads are added to building costs. Costs associated with
special unit or site-specific foundation requirements are not included and should be evaluated on a case-by-case basis.
9. To account for the cost of fittings, pipe lengths are determined by applying a multiplier to the overall system building layout
length. The resulting lengths are considered conservative (i.e., erring on the high side), so that the resulting cost covers
the installed cost of the pipe and fittings. The specific multipliers are as follows:
Combined influent and effluent pipe length is 2 times the length of the overall system building layout length.
Process pipe length is 2 times the length of the overall system building layout length.
Backwash pipe length is 2.5 times the overall system layout length.
Chemical piping length is 1 times the overall system layout length.

2.4.1 Building Costs

The WBS model building costs use three sources: RSMeans 2009 Square Foot Costs (RSMeans,
2008), Saylor 2009 Commercial Square Foot Building Costs (Saylor, 2009) and the Craftsman
2009 National Building Cost Estimator software model (described in Craftsman, 2008).
Appendix B provides a detailed description of these sources and the approach to developing
building costs.

In each WBS technology model, there are four possible design configurations for buildings: three
construction design and quality categories (low, medium and high) and small, very low cost,
prefabricated (“shed-type”). The WBS models select from among these configurations based on
system size, structure size and user input for component level (see Section 2.3.2), as shown in
Appendix B. Unit costs (in dollars per square foot) for each configuration vary by structure size.
When appropriate, the WBS models add costs for building heating and cooling systems as line
items separate from the base building costs. Whether the WBS models include these systems also
depends on system size, structure type and user input for component level, as shown in Appendix
B.

As discussed in Section 2.3.2, the input sheet of each model includes optional inputs that allow
the user to choose whether or not to include the costs of buildings and HVAC systems.

2.4.2 Residuals Management Costs

Many of the treatment technologies covered by the WBS-based models generate liquid, semi-
solid (sludge) and/or solid residuals. For these technologies, each model includes a sheet that
estimates the cost of various options for managing these residuals. The residuals management
options available for a given technology vary depending on the types of residuals generated, their

24
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

quantity, the frequency of generation (e.g., intermittent versus continuous) and their
characteristics. Examples of residuals management options include (but are not limited to): direct
discharge to surface water, discharge to a publicly owned treatment works, land disposal of
solids and storage and/or treatment of sludge or liquid waste prior to disposal or discharge. The
individual technology chapters of this document describe the specific residuals management
options available for each technology. Appendix C provides detailed information about the data
and assumptions used to estimate costs for the various residuals handling and disposal options.

2.4.3 Indirect Capital Costs

Indirect capital costs are costs that are not directly related to the treatment technology used or the
amount or quality of the treated water produced, but are associated with the construction and
installation of a treatment process and appurtenant water intake structures. Indirect costs can be
considerable and must be added to cost estimates if they are not included as a line item
component or a factor in the major (cost driver) elements of a technology. They include indirect
material costs (such as yard piping and wiring), indirect labor costs (such as process engineering)
and indirect burden expenses (such as administrative costs).

The WBS models compute the costs of site work, geotechnical investigation, yard piping and
standby power based on the system requirements, as determined during the direct capital cost
buildup. Other indirect costs are computed as a percentage of the installed process cost, building
cost or direct capital cost estimate. The indirect assumptions sheet in each WBS model (see
Section 2.3.5) contains guidance regarding a typical range of percentages for each item and
indicates the base cost to which the percentage will be applied. The guidance also describes
conditions that might require an assumption outside the range of typical values. Finally, guidance
on the output sheet notes that items such as installation costs and contractor overhead and profits
are already included in the direct capital cost estimate, but entries can be made to increase these
cost items should circumstances merit higher costs. Any of these costs can also be excluded by
modifying assumptions on the indirect assumptions sheet. Costs that are computed as a
percentage can be excluded simply by setting the percentage to zero. Those that are computed
based on system requirements can be included or excluded by setting the appropriate flag to one
or zero on the indirect assumptions sheet.

The WBS models report the total capital cost directly below this section of the output sheet so
the user can determine the impact of altering the indirect cost assumptions on total capital costs.

Appendix D provides descriptions of the default assumptions for the following indirect costs:

• Mobilization and demobilization

• Architectural fees for treatment building
• Equipment transportation, installation and contractor overhead and profit
• Construction management and general contractor overhead
• Process engineering
• Site work
• Yard piping
• Geotechnical
• Standby power

25
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• Yard wiring
• Instrumentation and control
• Contingency
• Financing during construction
• Legal, fiscal and administrative
• Sales tax
• City index
• Miscellaneous allowance.

2.4.4 Add-on Costs

Add-on costs are costs that may be attributed to one or more aspects of the treatment technology.
These add-on costs include permit costs (e.g., for construction and discharge permits), pilot and
bench testing costs and land use costs. Users can include or exclude these costs by setting
appropriate flags on the indirect assumptions sheet (see Section 2.3.5).

Permits
Systems installing new treatment technologies to comply with revised drinking water standards
will often need to build a new structure to house the new treatment train and might need to build
auxiliary structures to store chemicals (e.g., chlorine, which must be stored in a separate
building). In all jurisdictions, such construction activities require a building permit and
inspections to ensure that the structure meets local building codes. New treatment trains can also
create a new waste stream or supplement an existing one. New waste streams such as new point
source discharges to surface water generally require a state or federal permit; additions to
existing flows often require revisions to existing permits. The WBS models include costs for the
following permits:

• Building permits
• Permits under the National Pollutant Discharge Elimination System (when residuals
discharge to surface water is present)
• Storm water permits (for systems requiring one acre of land or greater)
• Risk management plans (when certain chemicals are present in large quantities)
• Compliance with the National Environmental Protection Act (included by default only at
the high cost component level – see Section 2.3.2).

Pilot Study
Site-specific pilot tests are often required by regulatory agencies to better define design
conditions and to ensure that the proposed technology will protect public health. In addition,
pilot tests and bench-scale tests can be run for non-regulatory reasons, e.g., to determine
appropriate loading and chemical feed rates, waste handling requirements or other process
parameters. Options for pre-design and pre-construction testing can include full- or small-scale
pilot studies, bench tests and desktop feasibility studies. Costs for pilot testing vary accordingly.

Pilot studies range from inexpensive small-scale efforts to full-scale tests that might be
warranted by site-specific conditions. Three variables affecting the costs of a pilot study are:
technology requirements, testing protocols and state requirements. Some states determine test

26
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

requirements on a case-by-case basis, particularly where drinking water standards or regulations

such as noise, air emissions, plume abatement or surface water discharges (e.g., the National
Pollutant Discharge Elimination System) are relatively stringent. The diversity of state
requirements, along with the many options for pre-design testing, means that requirements for
pilot- or bench-scale studies are difficult to define. Nevertheless, the WBS models include
default pilot study costs based on vendor quotes and estimated analysis costs. The user can alter
these costs by adjusting the permit cost data extracted from the central WBS cost database if site-
specific conditions warrant.

Land Cost
Regardless of whether a system needs to purchase additional land on which to build the new
treatment train, there is an opportunity cost associated with using land for water treatment rather
than an alternative use. The WBS models capture this cost in a land cost estimate that is based on
the calculated land requirement (in acres) and a unit cost per acre. As discussed in Section 2.3.2,
the input sheet of each model includes an optional input that allows the user to choose whether or
not to include the cost of land.

Each model estimates land required for the treatment system, plus a 40-foot buffer on one side
for emergency vehicle access and 10 feet on the other three sides. The user can change the
assumptions about buffer spacing using the critical design assumptions sheet for each
technology.

The WBS models incorporate land costs based on unit land costs that vary by system size and
land requirements that vary by technology and system size. Average land costs per acre are
estimated as probability-weighted averages using data from the Safe Drinking Water Information
System on system size and location, data for rural land costs for 50 states and data on urban land
costs for approximately 125 cities and metropolitan areas.

2.4.5 Annual O&M Costs

The O&M costs in each WBS model include annual expenses for:

• Labor to operate and maintain the new treatment equipment and buildings
• Chemicals and other expendable items (e.g., replacement media) required by the
treatment technology
• Materials needed to carry out maintenance on equipment and buildings
• Energy to operate all equipment and provide building heating, cooling, lighting and
ventilation
• Residuals discharge fees.

The individual technology chapters of this document describe additional, technology-specific

O&M costs.

O&M costs calculated in the models do not include annual costs for commercial liability
insurance, inspection fees, domestic waste disposal, property insurance and other miscellaneous
expenditures that are not directly related to the operation of the technology. These costs are

27
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

highly site-specific. Users wishing to include them should add the appropriate site-specific
estimates to the model results.

The WBS models calculate annual O&M costs based on the inputs provided by the user in the
input and O&M assumptions sheet. These inputs include system size, raw and finished water
quality parameters and other factors that affect operation requirements. Appendix E contains the
design assumptions used to develop default costs for the O&M sheet.

2.4.6 Total Annualized Cost

The output sheet in each model includes an estimated useful life, in years, for each WBS
component. The models take these component useful lives from the WBS cost database. The
useful lives vary by component type (e.g., buildings generally last longer than mechanical
equipment) and by material (e.g., steel tanks generally last longer than plastic tanks). The models
use the component useful lives to calculate an average useful life for the entire system. The
calculation uses a reciprocal weighted average approach, which is based on the relationship
between a component’s cost (C), its useful life (L) and its annual depreciation rate (A) under a
straight-line depreciation method. The formula below shows the reciprocal weighted average
calculation:
N

∑C n
C
Average Useful Life = n =1
N
=
∑A
A
n
n =1

where:
Cn denotes the cost of component n, n=1 to N
C denotes total cost of all N components
An denotes the annual depreciation for component n, which equals Cn/Ln
A denotes total annual depreciation for the N components.

The models use this average useful life for the system, along with a discount rate, to annualize
total capital cost, resulting in capital cost expressed in dollars per year. The models use a default
discount rate of 7 percent, which users can adjust directly on the output sheet. The models add
annual O&M cost to the annualized capital cost to arrive at a total annual cost in dollars per year.

2.4.7 Updating and Adjusting Costs

There are many factors that contribute to the variation in capital and O&M costs for the same
treatment technology. One variable is location, which is captured by the city index indirect cost.
Another is time—over time, the nominal price of materials, labor and land can change due to
inflation. If relative prices do not change over time (i.e., if innovative materials or production
technologies do not affect production cost relative to the price of other goods), then nominal
component prices can be adjusted using standard cost indices. The WBS cost database
incorporates the following indices to adjust prices to values in a common year:

• The Producer Price Index (PPI) consists of a family of indices that measure the average
trends in prices received by producers for their output (BLS, 2010). Within the PPI is the

28
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

family of commodity-based indices. The commodity classification structure of the PPI

organizes products by similarity of end use or material composition. Fifteen major
commodity groupings (at the two-digit level) make up the all-commodities index. Each
major commodity grouping includes (in descending order of aggregation) subgroups
(three-digit level), product classes (four-digit level), subproduct classes (six-digit level)
and individual items (eight-digit level). The WBS cost database assigns components to
the most closely related PPI commodity index. The selected price index for a component
is generally the index with the smallest product space. For example, prices for stainless
steel pressure vessels are escalated using a four-digit level index called BLS1072 Metal
Tanks.
• Building and construction costs are escalated using either the Engineering News-Record
Construction Cost Index or the Building Cost Index (ENR, 2013).
• Labor costs are escalated using the Employment Cost Index for “not seasonally adjusted,
total compensation, private industry and public utilities” (BLS, 2000; SIC series: 252).
The Bureau of Labor Statistics releases this index quarterly. The WBS cost database
utilizes an annual average.
• The Consumer Price Index is used to adjust land costs and components that have not been
assigned a specific PPI (BLS, 2007).

2.5 List of Abbreviations and Symbols in this Chapter

EBCT empty bed contact time
EPA U.S. Environmental Protection Agency
GAC granular activated carbon
gpm gallons per minute
GREPs generally recommended engineering practices
HVAC heating, ventilating and air conditioning
MGD million gallons per day
O&M operating and maintenance
P&ID process and instrumentation diagram
PPI Producer Price Index
TCE trichloroethylene
WBS work breakdown structure

2.6 References
American Water Works Association (AWWA). 1990. Water Quality and Treatment: A
Handbook of Community Water Supplies. Fourth Edition. New York: McGraw Hill.

AWWA/American Society of Civil Engineers (ASCE). 1998. Water Treatment Plant Design.
Third Edition. New York: McGraw-Hill.

Bureau of Labor Statistics (BLS). 2007. BLS Handbook of Methods: The Consumer Price Index.
Updated June 2007. Online at http://www.bls.gov/opub/hom/pdf/homch17.pdf

BLS. 2010. BLS Handbook of Methods: The Producer Price Index. Last updated 10 July. Online
at http://www.bls.gov/opub/hom/pdf/homch14.pdf

29
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

BLS. 2000. Employment Cost Indices, 1976-1999. September. Online at

http://www.bls.gov/ncs/ect/sp/ecbl0014.pdf

Craftsman Book Company. 2008. 2009 National Building Cost Manual. 33rd Edition. October.

Engineering News-Record (ENR). 2013. Building and Construction Cost Indexes. Online at
http://enr.construction.com/economics/

RSMeans. 2008. 2009 Square Foot Costs. 30th Annual Edition. Kingston, Massachusetts:
RSMeans Company.

Saylor Publications, Inc. 2009. 2009 Commercial Square Foot Building Costs: 19th Annual
Edition.

U.S. Environmental Protection Agency (U.S. EPA). 1997. Discussion Summary: EPA
Technology Design Workshop. Washington, D.C.: U.S. EPA, Office of Groundwater and
Drinking Water.

Viessman, W.J. and M.J. Hammer. 1993. Water Supply and Pollution Control. 5th Edition.
Harper Collins College Publishers, New York, NY.

30
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

3. Reverse Osmosis and Nanofiltration Model

Reverse osmosis (RO) is a membrane filtration technology that, in municipal drinking water
treatment, has traditionally been used for desalination of brackish waters and seawater. In RO,
influent (“feed”) water is passed through a membrane that filters out (or rejects) the substances to
be removed. Water passing through the membrane is split into two streams: the concentrate
stream, which contains the substances that are rejected by the membrane, and the permeate
(product) stream, which is the stream that passes through the membrane barrier (AWWA/ASCE,
2005). The process depends on applying high pressures across the membrane (in the range of
roughly 100 to 1,000 pounds per square inch gauge) in order to overcome the osmotic pressure
differential between the saline feed and product waters.

Nanofiltration (NF), sometimes called membrane softening, is a membrane filtration technology

that operates similarly to RO. NF removes ions that contribute to water hardness and other
contaminants of concern for drinking water treatment (AWWARF, 1995). The process uses
pressures in the range of 75 to 150 pounds per square inch (psi).

RO and NF membranes can remove materials as small as 0.01 to 0.001 microns. RO membranes
can remove inorganic ions and can effectively reduce total dissolved solids (TDS), nitrate,
radionuclides, total organic carbon, some disinfection byproduct (DBP) precursors and synthetic
organic chemicals. NF membranes efficiently remove divalent ions; they are less effective than
RO membranes in the removal of monovalent ions. They may therefore be used to remove
hardness. They are also used to remove color and odor compounds, synthetic organic chemicals
and some DBP precursors.

Exhibit 3-1 presents various particle sizes and shows the process range for RO and NF in
comparison to other membrane processes.

Exhibit 3-1. Particle Sizes and Membrane Process Ranges

Sources: after Osmonics (1990) and Dow (2005)

31
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Osmosis and reverse osmosis can take place when there is a concentrated solution on one side of
a semi-permeable membrane and a dilute solution on the other side. In the absence of any
externally applied pressure, water will flow through the membrane from the dilute side to the
concentrated side; this is osmosis. If, however, enough pressure is applied to the side with the
concentrated solution, the osmotic flow will stop; if more pressure is applied, the flow will
reverse and water will flow from the concentrated side to the dilute side. This condition is
reverse osmosis and the pressure that exactly balances the osmotic flow is called the osmotic
pressure.

RO is advantageous because of its capability to remove a wide range of contaminants. Since RO

permeate has a reduced chlorine demand, its finished water requires a low dose of disinfectant.
RO is suitable for small systems with a high degree of seasonal fluctuation in water demand.
There is no breakthrough period required for RO to work. NF and RO systems are automated
processes that require little operator time.

One of the disadvantages of RO is its low water recovery when compared to other drinking water
treatment technologies. The Small Business Advocacy Review Panel (1999) pointed out that a
water rejection rate of 20 to 25 percent can present a problem where water is scarce, such as in
the western states. Operating with a higher recovery, however, would increase equipment and
energy costs for small water systems. In addition, RO generates a large volume of concentrate
that may need to be treated before disposal. RO also has a higher energy demand than other
water treatment technologies due to its operation under high pressure.

3.1 Overview of the RO/NF Treatment Process

The RO/NF treatment process includes the following components:

• Feed water pumping

• Pre-treatment systems, which may include cartridge filters, pH control and antiscalant
addition, with secondary containment when chemical addition is included. (These
systems provide minimal pre-treatment that is usually appropriate for groundwater.
Surface water pre-treatment is considered in other work breakdown structure, or WBS,
models, as described below.)
• The RO or NF membrane filtration process system
• Chemical cleaning systems with secondary containment for cleaning chemical storage
• Equipment for the discharge of water treatment residuals
• Associated piping, valves and instrumentation.

Several processes that can be used for pre- or post-treatment in an RO/NF plant include the
following:

• When surface water is the source for an RO or NF membrane system, it is often

necessary to use other pre-treatment processes such as lime softening, media filtration or
low-pressure membrane filtration.

32
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• RO permeate has low alkalinity and pH and is therefore quite corrosive. Since gases
permeate freely through the membrane, it is typically high in dissolved carbon dioxide.
Hydrogen sulfide in the feed water will also be present in the permeate. Therefore, post-
treatment is necessary before the water can enter a distribution system. Typical post-
treatment steps include alkalinity recovery by feeding sodium hydroxide or lime and, if
needed, aeration for the removal of gases. Hydrogen sulfide may be removed by aeration
or oxidation with chlorine; in either case, it is necessary to maintain a low pH in order to
avoid the precipitation of elemental sulfur (AWWARF, 1996).

The WBS model for RO/NF does not include these types of pre- or post-treatment. To account
for the cost of these additional treatment steps, the U.S. Environmental Protection Agency (EPA)
has developed separate WBS models that generate costs for the pre- or post-treatment steps. In
generating national costs, EPA would add costs from the separate treatment models to costs from
the RO/NF model for scenarios that incorporate pre- or post-treatment.

Membranes can be made from organic and inorganic polymers and inorganic materials such as
ceramics or glass. The membranes used for water treatment are generally non-cellulose thin-film
composite (TFC) or amidic composite. Most drinking water systems use TFC membranes
because they can obtain higher production rates and better salt rejection rates than cellulose
membranes. Polyamidic and TFC membranes degrade in presence of chlorine and other oxidants.
They are resistant to biological fouling and hydrolysis between pH of 4 and 11 (AWWA, 1999;
AWWARF, 1996).

For municipal drinking water treatment, RO and NF membranes are typically used in a spiral-
wound configuration. A spiral wound membrane consists of several membrane envelopes,
layered with feed spacers and rolled together in a spiral around a central permeate collection
tube. Each envelope consists of a flat membrane sheet folded in half over a porous membrane
permeate carrier and glued on the remaining three sides to completely enclose the carrier. The
envelopes are connected to a central permeate collection tube. After the envelopes and feed
spacers are rolled around the tube, the assembly is enclosed in a shell to form a membrane
element.

Multiple elements are placed within a pressure vessel. To achieve the target removal efficiency
and water recovery, these pressure vessels often are arranged in sequential stages, typically up to
three depending on the recovery to be achieved (AWWA/ASCE, 2005; Dow, 2005). When
multiple stages are used, the number of pressure vessels decreases from stage to stage. Permeate
or finished water is collected from each pressure vessel. The concentrate stream from the first
membrane stage serves as the feed to the second and the concentrate stream from the second
stage serves as the feed to the third. Consequently, each successive stage of the process increases
the total system recovery (Jacangelo et al., 1998). The stages in combination make up an RO/NF
treatment train. A treatment system may have multiple trains. Exhibit 3-2 provides a schematic
drawing for the filtration process within an RO/NF system; each rectangular box within a train
represents a pressure vessel that contains multiple membrane elements.

33
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 3-2. Typical Schematic Layout for Reverse Osmosis/Nanofiltration

Membrane plant design generally requires a pilot study. One of the most important reasons for
conducting a pilot study is to evaluate the influence of water quality on membrane fouling and to
determine pre-treatment requirements. Calculations such as the silt density index (SDI), found in
ASTM standard D3739-94, can provide insight into the fouling problems that are inherent in any
membrane system. SDI measures the fouling potential of suspended solids. Manufacturers
typically specify maximum SDIs of 3 to 5 for RO and NF elements. In addition, it is important to
model and conduct pilot studies to assess the potential for fouling from substances such as
calcium carbonate (CaCO3), silica (SiO2), calcium fluoride (CaF2), barium sulfate (BaSO4),
calcium sulfate (CaSO4), strontium sulfate (SrSO4) and calcium phosphate. The scaling potential
of other substances may be determined from a saturation calculation. The Langelier saturation
index (LSI), described in ASTM standard D4189-94, characterizes the potential for CaCO3
scaling. The LSI is used to indicate whether calcium carbonate will precipitate or dissolve in
water, as well as the pH change required to re-dissolve the calcium carbonate.

Pre-treatment is necessary in a typical RO/NF system to minimize fouling and other damage to
the membrane. The type of pre-treatment necessary depends on the influent water quality, post
pre-treatment and recovery ratio and final water quality needed. The membrane material and
configuration of the RO/NF module will also affect the pre-treatment required (AWWARF,

34
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

1996). Typically, pre-treatment includes pre-filtration to remove large particles and addition of
acid and antiscalants to prevent scaling.

Periodic cleaning of the membrane system is necessary to recover productivity lost to fouling.
This cleaning may include cycles of acid and caustic wash, depending on the nature of the
fouling. Basic cleaning systems consist of chemical solution mixing tanks, screens and filters,
recirculation pumps, tubing and provisions for draining (Dow, 2005).

Different RO and NF membranes allow for different flux rates of water and ions. The passage of
water and ions may be modeled mathematically. There are many models of water and solute
transport across a membrane surface. The calculations in the WBS model are based on the linear
solution diffusion model (AWWARF, 1996).

Water flux, Kw, can be measured in gallons per square foot per day (gfd). It is characterized by
the equation:

𝐾𝐾𝑤𝑤 = 𝐴𝐴(∆P − ∆π)

where:
A = Water mass transfer coefficient or specific flux, gfd/psi
∆P = Transmembrane pressure differential, psi
∆π = Transmembrane osmotic pressure differential, psi

The quantity ∆P - ∆π is called the net driving pressure or NDP and is denoted Pnet.

Rather than measuring the solute flux directly, membrane system designers typically focus on the
salt passage, defined as:

𝐶𝐶𝑝𝑝
𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 =
𝐶𝐶𝑚𝑚

where:
Cp = Concentration on product side of membrane surface (permeate), mg/L
Cm = Concentration on feed side of membrane surface (concentrate), mg/L

Note that Cm is typically greater than the bulk concentration of solute on the feed side of the
membrane. This effect, called concentration polarization, is discussed further below. Typically,
each ion will have a different passage. The passage may be modeled for each ion in terms of
another membrane parameter, B, and a factor, f, that is specific to the ion-membrane
combination. Then:

𝐶𝐶𝑝𝑝 𝐵𝐵
= × 𝑓𝑓𝑖𝑖
𝐶𝐶𝑚𝑚 𝐴𝐴 × 𝑃𝑃𝑛𝑛𝑛𝑛𝑛𝑛 + 𝐵𝐵

where:
B = salt diffusion coefficient, defined for the membrane element, gfd
fi = Ion passage factor, dimensionless; varies for each ion, i, in the solute

35
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Typically, B is defined so that fi is equal to one for a reference ion such as sodium or chloride.

A and B depend on the membrane material and may change with the conditions under which the
membrane system is operated. They depend, in particular, on the water temperature, the pressure
applied to the membrane and the osmotic pressure and TDS on the feed side of the membrane
(Dow, 2005; Wolfe, 2004). As feed water flows through the membrane, ions are carried towards
the membrane surface. The ions that are rejected will concentrate near the surface, increasing the
feed concentration in a thin layer. This effect is known as concentration polarization. It is
characterized by the parameter β, defined such that:

𝐶𝐶𝑚𝑚
𝛽𝛽 =
𝐶𝐶𝑓𝑓

In this formula, Cm is the concentration at the membrane surface, as above, and Cf is the average
concentration in the feed water. The concentration polarization typically ranges from 1.0 to 1.2.
β increases with the recovery in the element and decreases with flow rate. According to Wolfe
(2004), it can be approximated by:

𝛽𝛽 = 𝑒𝑒𝑒𝑒𝑒𝑒𝑐𝑐𝑐𝑐

where:
Y is the fractional recovery in the membrane element
c is a coefficient that depends on the design of the element.

The recovery, Y, is equal to the fraction of feed water that passes through the membrane. The
default membranes in the WBS model have maximum recoveries ranging from 15 to 27 percent.
The recovery may be defined for a single membrane element or for the entire membrane system.

The pressure applied to the membrane determines the water flux; increasing the transmembrane
pressure will increase the flux through the membrane. At the same time, as water passes through
the membrane, the solute concentration on the feed or brine side of the membrane increases. The
designer must maintain a brine concentration low enough to prevent scale formation and fouling
of the membranes.

3.2 Input Sheet

The input sheet accepts user-defined design parameters that determine fundamental process
requirements. The user can indicate system size and select basic equipment parameters such as
operating pressure. Key design considerations that the user identifies on this sheet are described
in greater detail below and include the following:

• Feed water
• Production flows (see Section 2.3 and below)
• Membrane element
• Feed water temperature
• Include redundant train?
• Membrane element

36
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• Number of operating RO trains (optional)

• Target recovery rate
• Design flux
• Pre-treatment options
• Cleaning interval (optional)
• Residuals management
• Component level (optional, see Section 2.3)
• System automation (optional, see Section 2.3 and below)
• Include buildings; heating, ventilating and air conditioning (HVAC) and land? (optional,
see Section 2.3).

Exhibit 3-3 summarizes the units and constraints (i.e., Excel validation criteria) for each of these
inputs, along with conditions under which the model generates warnings. The sections below
describe each input in greater detail.

Exhibit 3-3. RO/NF Model Input Constraints and Warnings

Input Units Constraints Warning Conditions
Select Feed Water Pick list Pick list None
Design flow MGD or gpm Greater than 0 Blank
Blank, greater than design flow or
Average flow MGD or gpm Greater than 0
less than 0.1 times design flow
Blank or greater than the maximum
Feed temperature degrees centigrade Greater than 0 operating temperature specified for
the type of element selected
Include redundant train? Yes/No Yes/No None
Membrane element Pick list Pick list Blank
Number of operating RO trains
dimensionless Integer greater than 0 None
(optional)
Blank, outside constraints or cannot
Greater than 0 and
Target recovery rate percent achieve target recovery in three
less than 100%
stages
Blank or outside flux limits specified
Design flux gfd Greater than 0 on the critical design assumptions
sheet for the selected feed SDI
Acid for pretreatment Pick list Pick list Blank
Blank, less than minimum for
Treated feed pH (if acid pretreatment
Standard units Between 2 and 14 continuous operation or greater than
is included)
original feed pH
Antiscalant Pick list Pick list Blank
Greater than maximum allowable
Cleaning interval (optional) months Greater than 0 interval specified on the critical
design assumptions sheet
Discharge option for reject/cleaning
Pick list Pick list Blank
solution
Flow equalization for reject/cleaning
Pick list Pick list Blank
solution
Characteristics of spent
Pick list Pick list Blank
membranes/solids for disposal
Component level (optional) Pick list Pick list None
System automation (optional) Pick list Pick list None
Include buildings? Yes/No/Blank Yes/No/Blank None

37
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Input Units Constraints Warning Conditions

Include HVAC? Yes/No/Blank Yes/No/Blank None
Include land? Yes/No/Blank Yes/No/Blank None
gfd = gallons per square foot per day; gpm = gallons per minute; MGD = million gallons per day; SDI = silt density index

After the inputs are specified, the model will display a design for the RO system showing how
many stages are required, the number of pressure vessels per stage and how many elements will
be used in each pressure vessel. More information on the calculation is provided in Section 3.7.

To generate cost data, press the “Generate Results” button. The model will then determine the
required feed pressure and any pressure adjustments necessary in the later stages of the train, a
process that takes several seconds. During this time, the model will display messages that
describe the status of the calculations and may briefly show error messages that describe
problems with preliminary pressure estimates. These messages will disappear when the model
finds a feed pressure (and possibly a set of pressure adjustments in later stages) that satisfy the
target water production and recovery rate. In rare cases, the model may be unable to find an
appropriate set of pressures; in these cases, it may be necessary to adjust the target recovery rate
or design flux.

Note that the model’s designs are intended to generate cost estimates for policy analysis
purposes. Thus, the model estimates general quantity requirements for membrane elements,
pressure vessels, booster pumps or throttle valves and pump energy. The designs are not
necessarily optimized for the treatment of a particular feed water.

If you press the “Generate Results” button, adjust the inputs and press “Generate Results” again,
the model may display a dialog box saying that the main sizing inputs (design flow, temperature,
feed water, membrane element, target recovery and design flux) are the same as for the previous
run. In this case, it may be fastest to start from the pressures that the model computed for the
previous set of inputs. Choose “yes” at the dialog box to use these pressures. Advanced users
who change membrane element parameters or the critical design assumptions that govern the
pressure tuning routine, may wish to choose “no” to start from scratch. In either case, the model
should be able to compute a suitable set of pressures.

Once the model has completed its calculations, it will display information about the required
pressures, the permeate water quality and the scaling potential of the membrane concentrate
underneath the “Generate Results” button. If there is an error or other problem in the system
design, the model will display a message on the input sheet describing the problem.

The buttons at the top of the input sheet provide standard designs at several different flows. To
use the standard designs, first select a feed water and then press the button corresponding to the
desired flow.

38
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Feed Water

The WBS model for RO/NF includes a “drop-down” list box that allows the user to select from
among sever feed water options. This box is located at the top of the input sheet, above the
standard design buttons. The user should verify that the selection shown in this box is correct
before populating the other design input values. The user can change the feed water modeled by
picking a different selection from the list. After doing so, the user should then repopulate the
input sheet with values appropriate for the new feed water by clicking one of the standard design
buttons or manually adjusting inputs and clicking the “Generate Results” button (see Section 2.3
for further discussion of each of these methods).

The feed water options set the default feed water quality, which is a primary input to the train
design routine. Exhibit 3-4 shows the feed water options. EPA developed these water analyses to
establish typical low- and high-end costs for RO and NF water treatment, as well as typical mid-
range costs. Costs for specific scenarios may require changes in the feed water characteristics,
which require additional model inputs. 6 In addition to the options shown in Exhibit 3-4, a few
additional feed water designs are present for use in testing the model (e.g., 2,500 ppm NaCl). 7

The feed water also determines which membrane element and pre-treatment are used in the
standard designs (see Chapter 2 for more information on standard designs). The standard designs
for high, mid and low quality feed use nanofiltration, low-pressure RO and brackish water RO
elements, respectively. They also, by default, set the optional component level input to low, mid
or high cost, respectively (see Section 2.3 for more information on the component level input).
Each standard design includes chemical pre-treatment appropriate to prevent scaling given the
design’s feed water and recovery goal. A user can always override the standard designs by
entering different choices on the input sheet.

6
To use a different feed water, first enter it in the water specs sheet. An interactive form is available by using the
“Add Water Analysis” button on that sheet. The form allows users to enter the concentration of each ion. It will
issue a warning if the ionic charges do not balance. It is also possible to specify a feed water with a given
concentration of MgSO4 or NaCl. Each feed water must also be assigned a name, a pH, and an SDI. After the feed
water is entered in the Water Specs sheet, its name will appear in the Feed Water dropdown on the input sheet.
Choose its name from the dropdown. The model will then estimate costs for treatment of new feed water.
7
The model also includes a selection in the feed water list entitled “Reuse Train Post MF.” This option is meant to
be used in conjunction with other WBS models as part of a forthcoming tool to estimate the costs of a complete
series of treatment processes that might be used as part of an advanced water treatment train for potable reuse.
Because this tool is currently under development, this document does not discuss the specific inputs associated with
the reuse train feed water option.

39
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 3-4. Standard Feed Waters

High Quality Mid Quality Low Quality High Quality Mid Quality Low Quality
Water name
GW GW GW SW SW SW
Ca (ppm) 70 68 119 35 110 275
Mg (ppm) 18 50 100 5 40 25
Na (ppm) 47 300 1000 65 202 589
K (ppm) 3 0 20 1 7 3
Ba (ppm) 0.002 0.04 0.125 0.002 0.07 0.05
Sr (ppm) 1.1 0.3 10 1 1 1
HCO3 (ppm) 175 250 225 150 125 125
Cl (ppm) 40.5 60 1550 60 350 900
SO4 (ppm) 150 707 520 42 280 673
F (ppm) 0.1 1 2 0.1 0.3 0.3
SiO2 (ppm) 10 20 20 3 5 5
pH 7.3 7.6 8.0 7.5 7.9 8.2
SDI (15 min) 1.1 1.9 2.8 2.1 3.5 4.8
Approximate TDS (ppm) 510 1460 3570 290 1060 2530
Organics Organics
removal, TDS TDS removal, TDS TDS
Treatment goal
hardness reduction reduction hardness reduction reduction
reduction reduction
GW = groundwater; ppm = parts per million; SDI = silt density index; SW = surface water; TDS = total dissolved solids

Production Flows

As discussed in Section 2.3, the model requires the design and average flows to determine the
size and number of treatment components needed. In the RO/NF model, the input flows are the
design and average production, or treated water, flows. Unlike many other technologies, the
production flows in a membrane system will differ from the raw water, or influent, flows,
because recovery rates (see below under “Target Recovery Rate”) are less than 100 percent. The
WBS model takes into account recovery rates and sizes the equipment based on the raw water
flow needed to meet the production flow.

If the user specifies a bypass percentage on the critical design assumptions sheet (see Section
3.3), the production flows equal the total of the bypass and permeate flow. In this case the model
will compute the permeate flow required for the bypass percentage indicated by the user.
Otherwise, the production flow is the permeate flow. After the model determines the design and
average permeate flows, based on the specified production flows and any blending, it calculates
the necessary influent flow to achieve the desired production flow. It sizes the pumps and pre-
treatment components to handle the calculated influent flow.

Feed Temperature

The temperature of the feed water has a significant effect on the membrane’s permeability to
both solute and water (the water mass transfer coefficient A). The temperature also affects the

40
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

carbonate chemistry, osmotic pressure and viscosity. In general, a higher feed water temperature
will result in a higher permeate water flow, at a given pressure, as well as a higher salt passage.
The model will display a warning if the feed temperature is outside specifications for the selected
membrane element.

Systems that use surface water will experience seasonal variations in water temperature. As
stated above, the lower the temperature, the more pressure is required for a given amount of
permeate flow. However, since water demand also varies seasonally, it may be possible for
systems that use surface water to operate for fewer hours when water temperatures are low.
Instead of modeling temperature and demand throughout the year, the model uses an input for
design temperature (between 0 and 45°C) and computes pressure and energy requirements to
produce the average flow at the design temperature year round.

Include Redundant Train?

The input sheet offers the choice of whether or not to include redundant units in the system
design. The purpose of redundancy is to allow a plant to maintain its design capacity while some
membrane units are offline (e.g., for cleaning). Factors in determining whether to include
redundant membrane elements are the ratio of design flow to average flow, treated water storage
capacity, local regulatory requirements and other site-specific treatment goals or preferences.
Smaller systems with production flow rates less than 1 million gallons per day (MGD) often do
not include redundancy because their peak flow demands are much greater than their average
flow demands, creating a large excess production capacity on average. The standard designs
include redundancy for systems with design production flows greater than 1 MGD. The user can
modify the exact level of redundancy (i.e., the ratio of operating membrane elements to
redundant elements) on the critical design assumptions sheet.

Membrane Element

Membrane elements from different manufacturers, and different elements from the same
manufacturer, may have widely varying water and ion permeabilities. The WBS model contains
three basic types of membrane elements: low-pressure RO (with relatively higher water
permeability), brackish water RO (with lower water and salt permeability) and nanofiltration
elements. Each is available in three diameters: a 4-inch element (for use in very small systems),
an 8-inch element and a 16- to 18-inch element (the membrane industry does not currently have a
standard diameter larger than 8 inches, so a single large diameter [16 to 18 inches] is assumed in
the modeling). The element parameters in the model do not represent specific manufacturers.
Instead, they are chosen to represent mid-range values for A and B, based on the membrane

41
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

behavior modeled in design programs from different membrane element manufacturers.

Numerous characteristics of each element are tabulated on the ancillary sheet labeled
“Membrane Specs.” For each element, the following parameters are tabulated:

• Membrane type (RO or NF)

• Length and diameter of the element, in inches
• Membrane area, in square feet
• A value, in gfd/psi
• Coefficients used to correct A for applied pressure, temperature and feed osmotic
pressure
• B value, in gfd
• Coefficients used to correct B for temperature and feed TDS
• The geometric coefficient c used to compute the concentration polarization
• Coefficients used to compute the pressure drop as water flows through the membrane
element
• Maximum operating specifications, including the maximum feed pressure and flow,
maximum recovery in one element, minimum concentrate flow, maximum feed SDI,
minimum operating pH, maximum operating temperature and maximum pressure drop
permitted in a single element and a complete pressure vessel.

Section 3.7 details the use of these parameters.

The membrane element is selected using the drop-down box on the input sheet. When a user
chooses an element, the model displays its area. If the SDI of the feed water is too high for this
element, the model will display a warning. In this case, the user may choose another element or
create a customized feed water on the water specs sheet with an SDI level that reflects the
expected SDI level with pre-treatment of the feed water. The user may then use select the new
water as described above in the section “Feed Water.”

Number of Operating RO Trains (Optional)

By default, the model calculates the number of operating trains, before redundancy, based on a
maximum permeate flow per train that varies based on the membrane element specified, with a
minimum of two trains. The input sheet, however, includes an optional input that allows the user
to override the default calculation and directly specify the number of RO trains, before
redundancy, to be included in the design. The model standard designs leave this optional input
blank to accept the model default calculations.

42
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Target Recovery Rate

As discussed above, the RO/NF process produces a permeate flow (water with most dissolved
solids removed) and a concentrate flow (residual water rejected by the membrane). The recovery
is the percentage of the influent flow rate that is recovered as permeate rate. Increasing the
recovery will increase the concentration of dissolved solids in the membrane reject water and
will thus increase the required feed pressure and the potential for membrane scaling. Thus, the
achievable recovery depends on the quality of the source water as well as the pre-treatment of the
water (ASCE/AWWA, 1998), and systems with high levels of TDS in their feed water will
typically operate at lower recoveries than systems with lower levels.

For a given membrane and feed water, a higher recovery will require the use of more elements in
series. To account for the non-linearities associated with the vessels in series, the actual recovery
performance compared to the manufacturer’s specifications is accounted for by an efficiency
variable “Average recovery per element, as fraction of spec” that is specified as a percentage on
the Critical Design Assumptions sheet. The model accomplishes this by increasing the number
of elements per pressure vessel and/or by increasing the number of stages in the system.

For NF membrane elements, the target recovery will typically be between 80 and 90 percent. For
RO elements, the target recovery will typically be between 50 and 85 percent. The model
displays guidance with those bounds. However, systems with small production flows may need
very few elements in total to achieve their production and flux goals. Since these systems have
few elements in total, they cannot put many elements in series, and they may therefore be more
limited in the recovery they can achieve. The most economical choice for those systems may be
to run at a low recovery.

The model standard designs use target recovery rates between 75 and 85 percent when the design
flow is greater than 0.5 MGD. For design flows less than 0.5 mgd, the designs use recovery rates
that optimize the model’s projected costs. Recoveries at these small flows, therefore, vary
between 65 and 80 percent. In general, standard designs for lower cost levels use higher recovery
rates.

Design Flux

The flux of the system is the rate of permeate water per unit of membrane area. In the model, it is
measured in gfd. While each stage of a membrane system will have a different flux, the average
flux over all elements is a fundamental design parameter. In general, the higher the quality of the
feed water, the higher the flux that may be achieved. Operating with excessively high flux,
however, leads to fouling of the membrane elements. Depending on the nature of the fouling, it
can be reversed by cleaning or can require replacement of the elements.

For some high-quality groundwaters, systems can operate successfully with fluxes as high as 18
to 20 gfd. Surface waters require lower fluxes, typically between 12 and 17 gfd, depending on
the SDI of the source water. Pre-treatment with microfiltration or ultrafiltration membranes will

43
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

usually permit the use of a higher flux. The WBS model uses the SDI as a proxy for the feed
water quality. It will issue a warning if the flux is outside a prescribed range, based on the SDI
and whether the water is a surface or groundwater. The model’s flux ranges are based on typical
recommendations from membrane element manufacturers. Exhibit 3-5 lists these ranges.

Exhibit 3-5. Flux Ranges

Feed Water Type Feed Water SDI Flux Range
Either <1 a 21-25
Groundwater 1-3 16-20
Groundwater 3-5 12-16
Surface Water 1-3 13-17
Surface Water 3-5 10-16
a. These very low SDI values are typically associated with second-pass treatment of RO permeate.

The standard designs in the model use fluxes of 19, 17 and 16 gfd, respectively, for the low, mid
and high quality groundwaters and 16, 14 and 12 gfd, respectively, for the low, mid and high
quality surface waters.

Pre-treatment Options

To reduce fouling of the membrane, some type of pre-treatment is usually required. The RO/NF
model includes three pre-treatment steps, two of which are optional:

• 5-micron cartridge filtration, to remove larger particles that can clog the membrane
elements
• Optional acid addition, to reduce the scaling potential of calcium carbonate
• Optional antiscalant addition, to reduce scaling potential of several substances.

Other possible pre-treatment steps, such as lime softening, media filtration and low-pressure
membrane filtration, are covered by other WBS models and are not included in the RO/NF
model, as discussed in Section 3.1.

Reducing the feed water’s pH by adding acid will lower the concentration of carbonate ions and
therefore will reduce the saturation of calcium carbonate. Acidifying the water will also convert
some bicarbonate to carbonic acid, which passes through the membrane as carbon dioxide.
Adding lime to the permeate will then convert the carbonic acid back to bicarbonate, allowing
recovery of alkalinity in the finished water.

The model includes the option of adding hydrochloric acid or sulfuric acid; however, because
hydrochloric acid is much more expensive than sulfuric, the standard designs that include acid
pre-treatment always use sulfuric acid. If acid pre-treatment is specified, the model requires the
target pH for the feed water. It then determines how much of the specified acid to add to the

44
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

feed. The model will issue a warning if the feed pH is lower than the tolerance indicated for the
membrane element; however, elements can typically operate continuously with pH as low as 2 to
4, while a typical feed pH might be 5 to 6.5.

Antiscalants, also called scale inhibitors, will inhibit scaling when scaling salts are oversaturated
in the membrane concentrate. If the system design routine determines that salts are oversaturated,
the model will recommend the use of an antiscalant, unless the oversaturation is so great that
antiscalant is likely to be ineffective. Many different antiscalants are available, and the choice of
antiscalant depends on site-specific conditions. To capture some of this variation, the WBS
model includes two options for antiscalants: a basic antiscalant, for use when only carbonate
scaling is of concern, and a premium antiscalant, for use when silica, fluoride or sulfate scaling is
of concern. The WBS model uses the following rules of thumb (Freeman and Randtke, 2006) to
determine whether antiscalant will be effective:

• CaCO3, to an LSI of 2.3

• SiO2, to a concentration of 150 milligrams per liter (mg/L)
• CaF2, to 10,000 percent of saturation
• CaSO4, to 200 percent of saturation
• BaSO4, to 4,000 percent of saturation
• SrSO4, to 800 percent of saturation.

Users can adjust these parameters on the critical design assumptions sheet.

The input sheet contains a user input option that determines whether to use basic antiscalant,
premium antiscalant or neither. The antiscalant dose can be specified on the critical design
assumptions sheet.

A bicarbonate alkalinity of 1 to 3 milliequivalents per liter in the finished water is desirable for
corrosion control (AWWA, 1999). The standard designs in the model therefore use sulfuric acid
pre-treatment to ensure that at least 1.5 moles per liter of inorganic carbon is present in the RO
permeate.

Antiscalant use in the standard designs is based on the model’s projections of concentrate
saturation. The concentrate resulting from treatment of the low-cost waters is oversaturated in
calcium carbonate (LSI>0), while the mid- and high-cost waters yield concentrate that is also
oversaturated in other scaling salts. Therefore, for low and mid quality waters, the model’s
standard designs use premium antiscalant. For high quality waters, the standard designs use basic
antiscalant, except for the two smallest system sizes with high quality water, which use no
antiscalant.

Cleaning Interval (Optional)

By default, the model estimates a cleaning interval from 3 to 18 months, based on the SDI of the
feed water. However, cleaning requirements for RO/NF systems vary greatly based on site-

45
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

specific conditions. Users may therefore enter a cleaning interval on the input sheet, overriding
the model’s estimate. For more information on the estimate, see Section 3.3.

Residuals Management

The RO/NF process generates three residuals streams: the membrane concentrate, spent cleaning
solution and spent membranes. Since the spent cleaning solution is generated infrequently and in
small amounts, the model assumes that it will be diluted and discharged with membrane
concentrate. The input sheet requires the user to choose from among several options for
management of the resulting liquid residuals stream. Exhibit 3-6 shows the available residuals
management options for membrane concentrate (including diluted cleaning solution).

The management options shown in Exhibit 3-6 include several alternatives for flow
equalization. While membrane systems generate concentrate on a continuous basis, some
systems may operate for partial days. Flow equalization might be appropriate, then, to prevent
instantaneous flow from overwhelming the capacity of a publicly-owned treatment works
(POTW).

The option in the RO/NF model for direct discharge assumes that RO and NF facilities located
near a coastline are likely to use ocean outfalls to dispose of their concentrate. Therefore, when
direct discharge is chosen, the model assumes 5,000 feet of piping is required to reach the
discharge point, rather than the default value discussed in Appendix C.

Exhibit 3-6 also includes the option for an evaporation pond. Given the potentially large
quantities of membrane concentrate, this management method might be appropriate when the
residual is non-hazardous, particularly for facilities in dry climates. The required area for an
evaporation pond depends on local climate. After selecting an evaporation pond, the user should
carefully review the climatic parameters on the critical design assumptions sheet. For more
information, see Appendix C. Furthermore, to ensure that heating, cooling and standby power
costs are computed consistently, the user should review the climatic parameters in the operating
and maintenance (O&M) and indirect assumptions sheets. Appendix E and Appendix D contain
information on these parameters. The use of an evaporation pond results in the generation of a
secondary residuals stream, in the form of evaporation pond solids.

46
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit 3-6. Management Options for Membrane Concentrate

Options for Stage 1 of Residuals Management Options for Stage 2 of Residuals Management
Direct discharge to surface water or ocean outfall
Discharge to POTW
Holding tanks (for flow equalization) Deep well injection
Concentrate recycling *
Hazardous waste disposal
Direct discharge to surface water or ocean outfall
Discharge to POTW
Holding basins (for flow equalization) Deep well injection
Concentrate recycling *
Hazardous waste disposal
Direct discharge to surface water or ocean outfall
Discharge to POTW
Holding lagoons (for flow equalization) Deep well injection
Concentrate recycling *
Hazardous waste disposal
Direct discharge to surface water or ocean outfall
Discharge to POTW
No flow equalization
Deep well injection
Evaporation pond**
* This option has been temporarily removed from the model. The train design sheet does not take concentrate recycling into
account when it determines the feed pressure. See discussion below.
** Results in generation of secondary residuals (evaporation pond solids).

The model also includes the option of deep well injection of membrane concentrate. Larger RO
plants, notably in Florida, have injected their membrane concentrate into porous rock formations
below the lowest source of underground drinking water. The Class I injection wells used for
concentrate disposal can reach depths of 1,000 to 8,000 feet, depending on local geology. They
use several casings and layers of cement surrounding the central pipe through which concentrate
is injected, called the injection tubing. The cost of an injection well includes substantial testing
and surveying. A separate monitoring well is also required.

For deep well injection, the WBS model uses a linear approximation to a cost buildup developed
for the Bureau of Reclamation (Mickley, 2006):

Cost (2001$) = -288,000 + 145,900 * Diameter (inches) + 754 * Depth (feet)

The diameter in the approximation is the diameter of the central injection tubing. The cost
buildup includes costs for drilling and reaming; logging, testing and survey; installed casing,
grouting, injection tubing and packer; and a monitoring well. The approximation also includes an
allowance for mobilization and demobilization, equal to 20 percent of the construction costs for
the main well (not the monitoring well). It does not include the cost of pumping the concentrate
into the well.

To use the Bureau of Reclamation costs, the WBS model assumes a well depth of 3,500 feet. The
model will use multiple wells if the concentrate flow rate is greater than a maximum flow rate,
assumed to be 10,000 gallons per minute (gpm). The model determines the diameter of the
injection tubing by assuming a flow speed of 10 feet per second at the design flow of
concentrate. The model includes the mobilization and demobilization cost of the Bureau of

47
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Reclamation model in its indirect costs. However, for simplicity, the WBS model assumes that
one-sixth of the total injection well cost is due to mobilization and demobilization. The model
also includes pumps at the wellhead with 40 psi of head. When deep well injection is chosen, the
model assumes 5,000 feet of piping is required to reach the injection point, rather than the default
value discussed in Appendix C.

The model includes the option to recycle membrane concentrate to the head of the RO/NF
process. Concentrate recycling is sometimes used in RO/NF systems to improve crossflow
velocity and reduce the recovery required for the membrane system. It is important to note that
the model’s pressure calculations do not take concentrate recycling into account. To approximate
the effects of concentrate recycling, it is possible for users to enter a custom feed water with the
estimated characteristics of the combined feed and recycle streams and adjust the target recovery
downwards. This option has been temporarily removed from the model while EPA reviews its
implementation.

Finally, the model includes the option to dispose the membrane concentrate as liquid hazardous
waste. Flow equalization is always required when hazardous waste disposal is selected, to store
the membrane concentrate prior to disposal.

Management options available for spent membranes and evaporation pond solids (when
generated) include the following:

1. Disposal as non-hazardous solid waste 8

2. Disposal as hazardous waste
3. Disposal as radioactive waste
4. Disposal as radioactive and hazardous waste.
The solid residual management options do not include disposal in an on-site facility. This option
would be economically viable only for facilities with an existing on-site landfill—a factor that is
highly site-specific. For these facilities, the cost of this option would be less than that for off-site
disposal, because it would involve much lower transportation costs. Therefore, the off-site
disposal options included provide a conservative cost estimate for these facilities.

According to Burbano et al. (2006), the predominant method of concentrate disposal for RO/NF
plants is direct discharge to an ocean or receiving stream; 59 percent of plants surveyed used
direct discharge. Therefore, all of the standard designs in the model assume direct discharge of
the membrane concentrate to an ocean or surface water.

System Automation

8
Users can also select land application, instead of landfilling, for non-hazardous evaporation pond solids by
changing an option on the critical design assumptions sheet. Appendix C discusses this option in more detail.

48
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The input sheet includes an optional input for system automation. This input functions as
discussed in Section 2.3, except as follows. Membrane systems are typically sold with automatic
controls included, and the system automation input does not change the controls included in the
membrane system itself. Instead, it controls the automation of the rest of the plant and the
integration of the membrane system’s controls with the rest of the plant’s control system. If, for
instance, the user selects manual controls, the model will include time required for an operator to
operate pumps and valves for the pre-treatment and cleaning systems and manually record
instrument readings. However, the model will also assume that automatic controls accompanying
the membrane system actuate the valves found on the skids.

Since manual operation of a membrane facility is likely to be complex, the model assumes a
fully automated system regardless of component level or system size. This approach differs from
the default logic used in other WBS models and displayed in Exhibit 2-7. The model standard
designs, therefore, always include the costs of fully automated system controls.

3.3 Model Assumptions Sheets

There are three sheets that contain assumptions needed to facilitate process design: the critical
design assumptions sheet, the O&M assumptions sheet and the indirect assumptions sheet. These
sheets contain a variety of structural and chemical engineering parameters used in the
engineering design sheets.

There are more than 100 critical design assumptions in the model that cover process, O&M and
indirect cost parameters. Key critical design assumptions include pre-treatment chemical dosages
and characteristics, pre-treatment chemical storage, membrane fouling factor, interstage
throttling and pressure boost, membrane life, cleaning frequency, cleaning chemical dosages,
markup for rack assembly and bypass percentage. The following sections provide descriptions
and default values for these assumptions. Any assumption value can be modified, as needed.

Pre-treatment Chemical Dosages and Characteristics

Antiscalant dosage depends on feed water quality and water chemical composition. The model
computes an antiscalant dose in the feed water sufficient to provide 15 mg/L of antiscalant in the
concentrate (Watson, 2002). This concentration is specified on the critical design assumptions
sheet. To determine the antiscalant dose in the feed, the model multiplies the concentration by 1
minus the system’s recovery rate.

To calculate the quantity of chemicals consumed, the model also requires assumptions about the
density of acid and antiscalant. The model assumes the use of commercially available 93 percent
sulfuric acid or 28 percent hydrochloric acid, with densities of 14.19 pounds per gallon (lbs/gal)
or 2.662 lbs/gal, respectively. It assumes the use of 50 percent sodium hydroxide, with a density
of 6.364 lbs/gal. The concentrations can be adjusted on the critical design assumptions sheet.
Based on the concentrations, the model looks up the densities in the WBS cost database. The
model assumes an antiscalant density of 10 lbs/gal.

Pre-treatment Chemical Storage

The quantity of pre-treatment chemicals stored on-site determines the size of pre-treatment
chemical storage tanks and the initial quantity of pre-treatment chemicals purchased. According

49
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

to U.S. EPA (2000), tanks are generally sized to hold approximately 30 days worth of chemicals,
so the model uses this assumption to determine storage requirements.

Membrane Fouling Factor

The water permeability of a membrane element decreases as the membrane ages. The WBS
model uses the permeability specified for new membrane elements, but includes a fouling factor
to account for the decrease in permeability. For design purposes, the membrane’s water
permeability (its A value) is reduced according to this factor. The factor may be defined so that it
is compounded each year, modeling decreases in permeability as the membrane element ages, or
it may be defined so that it is applied only once during the design of a system. For simplicity, the
fouling factor in the WBS model is applied only once.

By default, the fouling factor is 85 percent, indicating that the permeability will be reduced by 15
percent. Reducing this factor will allow for more fouling and will therefore increase the pressure
required for a given flux and recovery.

Permeate Throttling and Interstage Boost

The train design sheet might determine that it is necessary to adjust the concentrate pressure
between stages. There are two ways to adjust the pressure: by using booster pumps before later
stages or by introducing a backpressure (throttle) on the permeate from earlier stages and
correspondingly increasing the overall feed pressure. While booster pumps increase the system’s
footprint and capital cost, they are generally a more energy-efficient option than permeate
throttling. Unless the throttle option is chosen on the Critical Design Assumptions sheet, the
model assumes the use of booster pumps when pressure adjustment is necessary.

Membrane Life
The life of the membrane elements determines their replacement frequency, which the model
uses to calculate operations and maintenance costs. According to the engineering experts who
reviewed the critical design assumptions, most engineers use a 5 year life when preparing cost
estimates, although actual life may be longer in practice. Consistent with this practice, the model
assumes a 5 year life for RO and NF elements (specified on O&M assumptions sheet).

Cleaning Frequency and Capacity

Because membranes are subject to fouling, periodic cleaning of the membrane system is
necessary. Indications that cleaning is necessary include: a pressure drop across the membrane,
an increase in the feed pressure requirement or a significant decrease in the permeate flow. The
model incorporates a cleaning regime where acid wash is followed by a high-pH wash. Cleaning
system design depends on several factors:

• Cleaning frequency (i.e., the number of months between cleanings)

• Cleaning cycle length (i.e., the number of days to complete the cleaning process).
In general, the cleaning frequency depends on the feed water quality and the system’s flux rate.
The WBS model permits the user to enter a cleaning interval directly in the input sheet, based on
site-specific conditions. If no cleaning interval is specified there, the model uses the feed water’s

50
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

SDI as a proxy for the water quality and estimates a cleaning frequency with the following rule
of thumb (Wolfe, 2004):

Interval between cleanings (months) = 12 – 4 * (SDI – 2)

When the feed water SDI is equal to 2, the model assumes 12 months between cleanings. When
the SDI increases by one, the model reduces the cleaning interval by 4 months, to a minimum
interval of 3 months. When the SDI decreases by one, the cleaning interval increases by 4
months, to a maximum interval of 18 months. The resulting cleaning frequencies agree with
estimated cleaning intervals provided by membrane equipment manufacturers. The parameters
for the cleaning interval estimate can be adjusted on the critical design assumptions sheet.

Cleaning Chemicals, Dosages and Equipment

Several different chemicals can be used in the cleaning process. Their dosages, along with
cleaning frequency and membrane system volume, determine the quantity of chemicals
consumed in the cleaning process. For the acid cleaning cycle, the model permits the use of citric
acid, hydrochloric acid or a proprietary acidic cleaner indicated for sulfate and metal fouling. For
the high-pH cycle the model permits the use of caustic soda alone or a high-pH detergent. The
model assumes the use of proprietary acidic cleaner and a high-pH detergent (Dow, 2005).

The model assumes the use of separate equipment for the acid and caustic cycles of cleaning. If
the user opts not to use an acid or caustic cleaning cycle, the model will adjust the amount of
capital equipment in the cleaning system accordingly. For more information on cleaning
equipment, see Section 3.9.

Markup for Assembly of Rack

The model estimates the cost for installation of the membrane system (assembly of the support)
rack using a percentage, specified on the critical design assumptions sheet, times the capital cost
of the membrane equipment, excluding elements and piping. The model assumes this installation
markup is 100 percent.

Bypass Percentage
Because RO and NF can reduce TDS or hardness to very low levels, systems may choose to treat
only a portion of their production flow when blending is allowed, using a smaller treatment
system and blending treated water with raw water while still achieving treatment targets. The
bypass percentage is that portion of production flow that goes untreated. If bypass is used, the
model designs the treatment system to treat a flow equal to (100 minus bypass percentage)
multiplied by design production flow and adds bypass piping and associated valves to the
components included on the output sheet. The RO/NF model also adds booster pumping for the
bypass line when bypass is specified. The model assumes no bypass, but the user can incorporate
bypass by entering a percentage of bypass flow on the critical design assumptions sheet.

3.4 Influent and Booster Pumping Sheet

The design of an influent pumping system is determined by the peak influent flow along with the
specific pumping head and pressure required for RO/NF treatment. The influent and booster
pumping sheet assumes as many influent pumps as required to handle the total feed flow into the

51
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

RO/NF process trains, without going above a maximum flow per pump, specified on the critical
design assumptions sheet. The pump number is determined by the number of pumps designed for
meeting the average flow rate necessary to operate at the maximum flow rate. The model
assumes no redundant pumps by default on the Critical Design Sheet. If a bypass percentage is
specified on the critical design assumptions sheet, this sheet also performs the required
calculations for the bypass pumps.

To determine the size of each pump, this sheet calculates the influent flow based on production
flow and recovery rate as specified on the input sheet. The sheet then increases this flow by a
safety factor specified on the critical design assumptions sheet and splits it among the operational
pumps. Finally, it computes the horsepower needed for each pump.

The model determines the costs of the appropriate horizontal split-case pumps based on the
required pressure and flow. It uses the horsepower to determine the pump footprint, by looking
up the required areas for pumping and electrical equipment in the WBS cost database. If booster
pumps are required, the model determines their size and energy requirements using the same
method as for the influent pumps.

3.5 Pre-treatment Sheet

As discussed above, RO/NF systems require pre-treatment facilities. For groundwater treatment,
pre-treatment typically includes filtration and chemical conditioning of the feed water. Surface
water requires further pre-treatment, possibly including lime softening, media filtration, low-
pressure membrane filtration or dissolved air flotation. As discussed above, these latter potential
pre-treatment steps are not included in the WBS model for RO/NF.

Cartridge filtration protects the membranes by removing large particles. The cartridge filters are
installed between the pretreatment system and the feed water pumps. For systems with
production flows below 1 MGD, the model uses one cartridge filter. From 1 to 5 MGD, the
model uses two cartridge filters. Above 5 MGD, the model will use one cartridge filter for every
5 MGD of design flow. In addition, the model includes one standby cartridge filter for systems
with design flows up to 50 MGD or two standby filters for plants with design flows of 50 MGD
and above. Users can adjust these assumptions on the critical design assumptions sheet. The unit
capital costs for the cartridge filters are based on manufacturer’s quotes and vary based on the
flow per filter. Each filter can contain multiple cartridges.

Addition of acid and antiscalant also can be necessary in a pre-treatment system, and the model
includes inputs to control whether these steps are included in the cost estimate. The pre-treatment
sheet determines acid and antiscalant doses based on the objectives set in the input and critical
design assumptions sheets, as described above. Equipment required for the acid and antiscalant
chemical systems includes the following:

• Tanks for acid storage

• Tanks for antiscalant storage
• Metering pumps for each chemical.

The model assumes separate day tanks, in addition to the primary bulk storage tanks, are
required for each chemical if daily chemical usage at design production flow exceeds a number

52
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

of gallons specified on the critical design assumptions sheet. To provide secondary containment
for the pre-treatment chemical tanks, this sheet also includes calculations for concrete curbing
around the storage area and chemical resistant coating for the curbing and the underlying
concrete pad.

3.6 Feed Water Sheet

The feed water sheet determines the ion concentrations in the feed water, based on the analysis
of the feed water selected by the user. Given the ion concentrations and pH listed for the selected
feed water, the sheet computes the concentrations of carbonate ions and of carbon dioxide and
makes sure the anions and cations in the water are balanced. The model then computes the LSI
and the saturation percentages of SiO2, CaSO4 and CaF2 in the feed water, for display on the
input sheet. The methods used to compute these quantities come from the Total Flux and Scaling
Program (TFSP), developed by PerLorica, Inc. for EPA.

If the user specifies acid addition, the feed water sheet determines the concentrations of
carbonate ions and carbon dioxide for the adjusted feed pH. It then determines the required
addition of chloride or sulfate ions by requiring that the treated feed be charge-balanced and
determines the required dose of hydrochloric or sulfuric acid in commercial concentration to
achieve the target pH.

Finally, the sheet assembles a water analysis (viscosity and ionic concentrations) for the treated
feed and determines scaling quantities, as above. The water analysis is passed to the train design
sheet.

3.7 Train Design Sheets

3.7.1 Overview
The RO/NF model determines a train design based on the feed water, membrane element and
flux and recovery targets set on the input sheet. It operates as follows:

Based on the flux and recovery targets, the model determines how many total membrane
elements are needed and how they should be assembled into a train (that is, how many stages to
use, how many elements to include in a pressure vessel and how many pressure vessels to use in
each stage). These results determine most of the capital cost of the membrane system, except for
possible costs associated with booster pumps or permeate throttle valves. These computations
take place on the train design sheet, based on additional calculations that cover three additional
sheets.

Based on the feed water analysis and membrane element parameters, the model determines what
feed pressure is required to achieve the target recovery, thereby establishing pumping energy
requirements. If necessary, the model will include booster pumps or permeate throttle valves in
this step.

The model uses an iterative procedure to choose the feed pressure. Starting from a pressure
guess, it determines whether the membrane system would operate within specs and computes the
amount of permeate produced. It then modifies the pressure guess if it finds off-spec operation or
if the resulting permeate production is too high or too low. The iteration continues until the

53
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

model finds a feed pressure and interstage pressure adjustments that will yield the desired
permeate flow.

These computations take place on the pressure tuning sheet. They are controlled by a Visual
Basic macro, which runs automatically when the user presses the “Generate Results” button on
the input sheet.

To perform each iteration on the pressure tuning sheet, the model does a detailed, element-by-
element simulation of the membrane system. The simulation iteratively determines concentrate
and permeate flows and ion passages, based on the selected membrane element parameters and
feed water analysis. All calculations take place on the calc system sheet and calc one element
sheets. The iteration is controlled by a set of Visual Basic macros, which are run automatically
from the pressure tuning macro.

Thus, the train design process includes the four sheets labeled train design, pressure tuning, calc
system and calc one element. The following subsections describe how each sheet operates.

3.7.2 Train Design Sheet

The train design sheet uses the required flux and recovery to assemble a train configuration
(number of stages, number of elements per pressure vessel and the number of pressure vessels in
each stage). It also makes a first guess for the required feed pressure, which is used as an input to
the pressure tuning routine. The sheet uses a modified version of the design procedure outlined in
the Dow Filmtec technical manual (Dow, 2005).

The sheet starts by determining how many elements to use in series across all stages of the
system. It first computes the average recovery desired in each element, by taking 70 percent of
the maximum recovery specified for the element. (Users can adjust the fraction of the maximum
recovery on the critical design assumptions sheet.) The system recovery Ys is equal to

𝑌𝑌𝑠𝑠 = 1 − (1 − 𝑌𝑌𝑒𝑒,𝑎𝑎𝑎𝑎𝑎𝑎 )𝑁𝑁𝑒𝑒 ,

where Ye,avg is the average recovery per element, and Ne is the number of elements in series.
Therefore,
𝑙𝑙𝑙𝑙(1−𝑌𝑌𝑠𝑠 )
𝑁𝑁𝑒𝑒 = .
𝑙𝑙𝑙𝑙(1−𝑌𝑌𝑒𝑒 )

If the number of elements in series is less than 7, the model uses a single stage with the required
number of elements in each pressure vessel. If the number is greater than 7, it will use multiple
stages. It will use no fewer than 6 elements per vessel in this case and will choose the number of
elements per vessel and number of stages to come as close as possible to the desired number of
elements in series.

Users can adjust the minimum and maximum number of elements per vessel on the critical
design assumptions sheet. In particular, users who want designs with no more than 6 elements
per pressure vessel may change the assumption el_per_vess_max_8 to 6. Elements with a
diameter of 16 to 18 inches use different assumptions, with a minimum of 4 elements per vessel
in multi-stage systems and a maximum of 5 elements per vessel.

54
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The train design sheet then determines the ratio of pressure vessels in each stage. It attempts to
design the system so that the recovery can be the same in each stage. It follows (Dow, 2005) that
the ideal ratio of pressure vessels from one stage to the next is equal to

𝑅𝑅 = (1 − 𝑌𝑌𝑠𝑠 )−1/𝑛𝑛 ,

where R is the staging ratio, and n is the number of stages.

Based on the target flux and the membrane area in each element, the sheet determines the total
number of membrane elements and pressure vessels required. It arranges them in stages
according to the staging ratio it computes. It estimates the flow through each stage by assuming a
constant recovery in each element. Based on these flows, it uses membrane element
specifications such as the maximum recovery per element, maximum pressure drops, maximum
feed flow and minimum concentrate flow to constrain the number of vessels in each stage. (Note
that the flows are computed in detail later in the train design sheet; the flow estimate is used only
to adjust the train design.)

Finally, the sheet assembles a first guess for the feed pressure. It estimates the net driving
pressure, based on the target flux and an estimated A-value for the membrane; the average
osmotic pressure, based on the feed osmotic pressure, the target recovery and with an assumption
of 100 percent salt rejection; and the pressure drop in the system, using the flow estimate
described above. The feed pressure guess is the sum of these:

𝑃𝑃𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 = 𝑃𝑃𝑛𝑛𝑛𝑛𝑛𝑛,𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 + 𝜋𝜋𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 + 𝛿𝛿𝛿𝛿𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 /2.

3.7.3 Pressure Tuning Sheet

The pressure tuning sheet, along with a set of Excel macros, refines the first pressure guess until
it finds a feed pressure that will achieve the required flow of product water. It will adjust
pressure in the later stages if necessary. It uses the following steps:

• Load the pressure guess from the previous iteration or from the train design sheet on the
first iteration.
• Determine the flow and quality of the resulting permeate water or whether the pressure
guess would result in off-spec operation of the membranes. This step takes place largely
on the calc system and calc one element sheets.
• If the pressure guess results in off-spec operation, adjust the pressure to bring operation
within spec. The correction is estimated on the calc one element sheet.
• If the pressure guess results in a flux imbalance between stages, adjust the pressures to
rectify the imbalance. The model will adjust the pressures if the flux in any stage is more
than 25 percent greater than the required average flux. For the first stage, the model will
adjust the pressures if the flux is lower than the average flux for the entire system. For
later stages, it will adjust pressures if the flux is more than 25 percent lower than the
required average flux. Users can adjust these parameters on the critical design
assumptions sheet.

55
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• If the elements are operating within spec and the flux is sufficiently balanced, the model
compares the resulting permeate flow with the required flow. It then adjusts the pressure
proportionately to the required change in the flow.
• The sheet stops iterating when the permeate flow is within 0.5 percent or 500 gallons per
day (gal/day) of the required permeate flow or after 20 iterations. Users can also adjust
these parameters on the critical design assumptions sheet.

3.7.4 Calc System and Calc One Element Sheets

For each pressure guess, the model determines the resulting product water as follows:

• Load the inputs collected in the calc system sheet.

• Copy the element parameters into the calc one element sheet.
• Copy the initial feed water parameters into the calc one element sheet.
• At the beginning of a stage, copy the throttle and boost pressures into the calc one
element sheet, as required.
• Simulate the performance of the first element. This calculation is described below.
• If the calculation of the element’s performance results in an error, the calculation stops.
• Copy the results from the first element into the calc system sheet.
• Take the ion concentrations from the first element and copy them into the feed
concentrations for the second element.
• Repeat the previous four steps for each element in the first stage, up to the number of
elements per pressure vessel specified on the input sheet.
• Compute the recovery and other quantities for the entire first stage. These quantities are
described below.
• The process repeats for the second and third stages, if the train design sheet determined
that later stages are necessary.

Input Collection
At the top of the calc system sheet, the model displays the detailed parameters for the membrane
element chosen on the input sheet. Beneath them are two inputs that control the iteration process
for a membrane element. It should not be necessary to change these values.

Detailed Output
The system outputs are largely the same as those summarized on the input sheet. Other system
outputs here include the concentrate flow, TDS and the concentration of each ion in the permeate
water.

The outputs per stage include flows, recovery and permeate and concentrate TDS for each stage.
In this section, the information on recovery applies only to the given stage. For instance, a
system might have 48 percent recovery in the first stage and 35 percent recovery in the second
stage, for a total recovery of 66 percent. Likewise, the information on TDS applies only to the

56
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

given stage. For example, a system might have TDS of 19 in the product water from the first
stage and 61 in the product water from the second stage, for a final TDS of 31.

The outputs per element consist of a detailed table of results for each element in the train,
including flows, pressures and water analyses. More information on these results follows below.

Element Calculations
The detailed calculation of flow for each element takes place on the calc one element sheet. This
sheet contains data for only one element at a time; the data is loaded by the train design macro,
as described above. At the top of the sheet are the inputs, describing the selected membrane
element, the water fed into the element and the feed flow and pressure. A user may, if he or she
wishes, enter values into the bold gold cells by hand and then press the “Solve Element” button
to see the computation of the recovery and salt passage for the element. Changing values on this
sheet will not affect the cost estimate on the output sheet.

The element calculation computes the recovery and salt passage by determining the net driving
pressure and the membrane's A (water permeability) and B (salt permeability) values. The net
driving pressure depends on the osmotic pressure on the concentrate side of the membrane, as
well as on the feed pressure drop. The pressure drop, in turn, depends on the average of the feed
and concentrate flows. The A and B values also depend on the ions present in the feed and
concentrate water. A guess for the net driving pressure and the A and B values allows the model
to compute the water and salt passage for the membrane. The water and salt passage, in turn,
allow the model to compute the flow and ion concentrations on the concentrate side of the
membrane, which are used to recompute the net driving pressure and A and B values. The sheet
aims to find self-consistent values of the net driving pressure and of A and B. It proceeds as
follows:

• Convert the inputs into standard units and convert feed ion concentrations into
milliequivalents per liter.
• Corrects A and B for the temperature of the feed water and for the feed pressure.
• Establish first guesses for A and B, based on the osmotic pressure and TDS of the feed
water.
• Establish a first guess for the net driving pressure, based on the feed water and the salt
passage as determined by A and B
• Compute permeate and concentrate flows.
• Compute the concentration polarization, based on recovery.
• Compute the passage of each ion into the permeate, making sure the ions are balanced.
• Based on the ions in the permeate, determine the ions in the concentrate.
• Compute the osmotic pressure difference between the permeate and concentrate.
• Compute the pressure drop in the vessel, based on the feed and concentrate flows and
viscosities.

57
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• Recompute the net driving pressure, based on the osmotic pressures and the pressure
drop.
• Recompute A and B, based on the osmotic pressure and TDS in the brine, respectively.
• Compare net driving pressure, A and B to their values at the beginning of this iteration. If
they all agree within the tolerance specified on the system calculation sheet (0.01 percent,
by default), the computation is complete. If not, the new values are used to drive another
iteration. The model will perform up to 100 iterations.

On each iteration, the model corrects A and B to account for the effect of applied pressure, feed
osmotic pressure, feed salinity and temperature. The correction factors for A are as follows:

• Temperature correction factor,

1 1
exp[𝐾𝐾𝐴𝐴𝐴𝐴 ( − )],
𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟 𝑇𝑇
where kAT and Tref are specified for each membrane element. The model includes two
values of kAT for each element, one used when T>Tref and one used when T<Tref. For
some elements, these values are equal.
• Applied pressure correction factor,
1 − 𝑘𝑘𝐴𝐴𝐴𝐴 (𝑃𝑃𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 − 𝑃𝑃𝑡𝑡ℎ𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 ),
where kAP is specified for each membrane element.
• Osmotic pressure correction factor,
1 − 𝑘𝑘𝐴𝐴𝐴𝐴 𝜋𝜋,
where kAπ is specified for each membrane element, and π is the average osmotic pressure
on the feed side of the membrane.

Each correction factor is computed and applied to A independently of the others.

The correction factors for B are as follows:

• Temperature correction factor,

1 1
exp[𝐾𝐾𝐵𝐵𝐵𝐵 ( − )],
𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟 𝑇𝑇
where kBT and Tref are specified for each membrane element. The model includes one
temperature coefficient for B.
• Salinity correction factor,
1 + 𝑘𝑘𝐵𝐵𝐵𝐵 |𝑇𝑇𝑇𝑇𝑇𝑇 − 𝑇𝑇𝑇𝑇𝑇𝑇min 𝐵𝐵 |,
where TDS is the average TDS on the feed side of the membrane, and the vertical bars
denote absolute value (so that the factor is always greater than or equal to 1).

The B value of a membrane tends to be large at very low TDS, decrease to some minimum value,
and then increase again (Franks, 2005; Wolfe, 2006). For each element, the model includes the
minimum B value and the TDS min B at which it occurs. It also includes two values of the slope
kBS, one for use when TDS<TDS min B and one for use when TDS>TDS min B. Finally, the salinity
correction factor will not increase above a maximum value, also tabulated for each element.

58
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The correction factors are computed and applied to B independently of one another.

The model computes the pressure drop in each membrane element according to a power law with
a correction for viscosity,

𝛿𝛿𝛿𝛿 = 𝑘𝑘𝑃𝑃 𝑄𝑄𝑎𝑎 (µ/µ𝑟𝑟𝑟𝑟𝑟𝑟 )1/2 ,

where Q is the average of the feed flow into the element and the concentrate flow out of it, µ is
the average viscosity of the feed water (computed from the temperature and TDS), and µref is the
viscosity for the element’s reference temperature and TDS. The coefficient kP, exponent α, and
reference temperature and TDS are specified for each element.

After the train design macro runs, the model displays the computation for the last element in the
last stage of the system on the calc one element sheet. To view the computation for a different
element in the train, go to the train design sheet and select the cell containing the code for the
selected element. The codes are located at the top of the element output table. For instance, the
code for the fourth element in stage 1 is el_1_4. After selecting the code, press the button labeled
“Load Element Data.” The model will load the data for that element onto the element
calculation sheet and then will redo the calculation for that element.

3.8 Membrane Capital Sheet

The membrane capital sheet uses the results of the train design (see Section 3.7) to determine
how many membrane racks are required. The model assumes that each rack operates as a
separate process train. Racks are built in various sizes to house multiple pairs of pressure vessels,
and the size of a rack for a particular flow train will vary with flow rate, water quality, target
recovery, membrane element size and even design feed water temperature. The number of racks
affects cleaning requirements, the footprint required for membrane equipment, flow rates and
pipe sizes to and from each rack and instrumentation requirements.

The model first adds redundant pressure vessels, if redundancy is specified on the input sheet. By
default, it adds one redundant vessel for every ten operating vessels; users can adjust this
redundancy frequency on the critical design assumptions sheet.

The critical design assumptions sheet specifies the maximum permeate flow rate per rack, based
on the membrane element sizes (4-inch, 8-inch or 16-18-inch). The total number of racks is
determined by dividing the design permeate flow rate by the maximum permeate flow rate per
rack.

The cost of the RO/NF membrane system is determined for a package, including racks, pressure
vessels, membrane elements, valves and piping on the racks and controls for the equipment on
the rack. (Note that the membrane system component cost includes a concentrate control valve,
which is displayed in Exhibit 3-2 because of its importance in process control.) The cost is based
on quotes from membrane system integrators and is parameterized in terms of the total
membrane area supported. The costs do not include pumps, instruments or ancillary equipment
such as cleaning and pre-treatment equipment; costs for these items are included separately on
the output sheet.

59
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The capital cost of the initial set of membrane elements is included in the membrane system cost.
However, the membrane capital sheet also determines a replacement cost for membrane
elements. The replacement cost is based on a cost per unit area tabulated in the centralized cost
database and the area for each element tabulated on the membrane specs sheet (see Section 3.16).

3.9 Cleaning Sheet

The membranes need to be cleaned when, for example, the pressure drop across the membrane
increases, feed pressure requirements increase or product flow decreases beyond desirable points
or fouling or scaling is evident. Chemical cleaning for RO/NF systems may consist of cycles of
acid washing, caustic washing or washing with other chemicals; each cycle typically includes a
soak, a period of high-flow pumping of cleaning solution through the element, a flush with
permeate water and a period of filtering feed water to waste to ensure that the membranes are
performing as desired (Dow, 2005). The WBS model assumes that cleaning consists of an acid
cycle and a caustic cycle. The cleaning frequency is determined as described in Sections 3.2 and
3.3.

Exhibit 3-2, above, includes the layout of a typical chemical cleaning system for RO. The
cleaning equipment includes the following for each chemical:

• Tanks
• Heaters
• Cleaning pumps
• 100-mesh security screens to protect the pumps
• Cartridge filters to protect the membrane elements
• Associated valves and controls.

Typically, one stage of a multi-stage membrane system is cleaned at a time. The model therefore
assumes that the cleaning system is sized to wash all the first-stage pressure vessels on a single
rack. It sizes the tanks to accommodate a volume of cleaning solution equal to the total volume
of the vessels to be washed, plus an allowance of 5 percent for the volume of piping between the
cleaning system and the membrane rack. Pumps, screens and filters are sized for the high-rate
flow phase (specified on the critical design assumptions sheet); the flow is determined based on
the membrane element diameter.

The model sizes immersion heaters to heat the cleaning solution to 95 degrees in five hours,
starting at the feed water temperature. It uses the following rule of thumb (Cole-Parmer, 2006) to
determine the required heater power:

𝑉𝑉 × 𝛥𝛥𝛥𝛥
𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 =
325 × 𝑡𝑡

where the tank volume V is in gallons, the temperature increase ∆T is in degrees Fahrenheit, and
the heat-up time t is in hours.

Most facilities will include only one cleaning system, with some redundant equipment. Very
large facilities may need to perform simultaneous cleanings, depending on the number of racks,
the cleaning interval and the amount of time each rack is offline for cleaning. The model

60
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

assumes that each rack will be offline for two days to be cleaned and includes an additional
cleaning system if simultaneous cleanings are possible.

Finally, the model determines the cleaning chemical quantities (citric acid, hydrochloric acid,
caustic soda or proprietary cleaners) used for each cleaning, how much of each chemical must be
stored and the annual use of each cleaning chemical. If it is necessary to store more than a
minimum number of gallons (specified on the critical design assumptions sheet) of a cleaning
chemical, the model includes a storage tank in the cost buildup. To provide secondary
containment for the cleaning chemical storage tanks, this sheet also includes calculations for
concrete curbing around the storage area and chemical resistant coating for the curbing and the
underlying concrete pad.

3.10 Pipes and Structure Sheet

The pipes and structure sheet performs calculations for the following pipes:

• Influent and effluent piping

• Feed piping
• Permeate piping
• Cleaning system piping
• Bypass piping.

Calculations for the concentrate and residuals piping are covered on the residual management
sheet.

The size (diameter) of pipes is determined using a pipe flow look-up table that is part of the
WBS cost database. The pipe diameter selection method incorporates a reasonable head loss and
flow velocity, as documented in Exhibit 2-8. These design assumptions may result in some over
sizing of pipes, which means the costs for pipes may be conservative (i.e., err on the high side).

The flow used to determine influent and effluent pipe size is the design flow. The diameter of
interconnecting feed piping uses the same pipe flow chart, after splitting the inflow by the
number of skids; the diameter of permeate piping is determined by splitting the permeate flow by
the number of skids. Cleaning system piping is sized for the high-rate flow phase of cleaning,
using the same table. A similar approach is used in determining the size and capacity of bypass
piping. Except for cleaning system piping, the length of these pipes is determined using the
assumptions documented in Exhibit 2-10, which are designed to account for the cost of fittings.
For cleaning system piping, the model uses a length equal to the overall system building layout
length.

This sheet also calculates the housing area for RO/NF based on the footprint of the technology
components footprint and the spacing criteria specified on the critical design assumption sheet.
The space requirements for pumps, tanks, screens, filters and service space are based on
manufacturer specification, “to scale” drawings and the experience of engineers. The amount of
additional concrete needed to support heavy equipment, such as pumps and vessels, is calculated
using the footprint of the vessels and pumps.

61
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Square footage assumptions specified on the critical design assumptions sheet determine how
many buildings are required. For smaller space requirements, the model assumes a single
building containing all process equipment. For medium requirements, the model assumes two
buildings, one containing influent pumping and pre-treatment equipment and one containing
membrane equipment, cleaning systems and residuals management equipment. For larger
requirements, the model assumes three buildings, one containing influent pumping equipment,
one containing pre-treatment equipment and one containing membrane equipment, cleaning
systems and residuals management equipment. The number of buildings affects the total land
required and energy costs for heating, ventilating and cooling.

3.11 Instrumentation and Control Sheet

Controls for the valves on the membrane skids are integrated into the membrane system
component; unit costs in the WBS cost database include these controls. Therefore, the
instrumentation and control sheet calculates the incremental requirements for valves,
instrumentation (e.g., flow meters) and automated system controls that are external to the skids.
These include instrumentation and controls associated with pumps, pre-treatment and cleaning
and valves and controls associated with external piping. The number of valves and instruments is
based on the number of process components (e.g., number of feed water pumps) and assumptions
from the critical design assumptions sheet (e.g., number of valves per feed pump). The
assumptions correspond to the general schematic layout for this technology shown in Exhibit
3-2. Sizing of valves corresponds to the size of the appropriate pipe determined on the pipes and
structure sheet, except for the sizing of permeate throttle valves; these valves are sized according
to the flow rate from one skid due to their stage. Appendix A describes the method used in the
WBS models to estimate the number and type of system control components.

3.12 Residuals Management Sheet

The residuals management sheet estimates the volume and mass of residuals, their characteristics
and the capital and O&M requirements for residuals management, based on the management
options selected on the input sheet and the approach outlined in Appendix C. Depending on the
management options chosen, specific items of capital equipment for residuals may include:

• Holding tanks, basins or lagoons

• Pumps
• Evaporation ponds (including excavation and lining)
• Deep injection wells (with costs determined as outlined in Section 3.2)
• Coagulant feed and mixing equipment
• Valves, piping and instrumentation.

Specific O&M requirements associated with residuals may include:

• Residuals pump labor, materials and energy

• POTW discharge fees
• Coagulant usage

62
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• Membrane concentrate transportation and disposal costs (when hazardous waste disposal
is required)
• Used cartridge filter and membrane element transportation and disposal costs
• Holding tank or evaporation pond solids transportation and disposal costs.

The option for concentrate recycling appears in the menu but is not yet an available option. This
option would be for regions where high overall recoveries are necessary to meet water demands.

3.13 O&M and HVAC Sheets

The O&M and heating, ventilating and air conditioning (HVAC) calculations cover two sheets:
the O&M sheet (annual labor, materials and energy usage) and the HVAC sheet (HVAC capacity
requirements). The O&M sheet derives annual O&M requirements for RO/NF treatment based
on the engineering design, O&M critical design assumptions and input values. It determines the
following O&M requirements based on the approach outlined in Section 2.4 and Appendix E:

• Operator labor for system operation and maintenance

• Managerial and clerical labor
• Influent and booster pump maintenance materials and operating energy
• Facility maintenance materials
• Energy for building lighting and HVAC.

In addition, the sheet adds the following technology-specific O&M requirements:

• Labor for pre-treatment, the membrane process system and cleaning

• Maintenance materials for pre-treatment, membrane process and cleaning equipment
• Replacement cartridge filters and membrane elements (note that disposal of these items is
calculated on the residuals management sheet)
• Chemical usage (acid and antiscalant for pre-treatment and hydrochloric acid and caustic
for cleaning)
• Maintenance labor and materials and operating energy for bypass booster pumps (if
bypass is specified).

The sheet determines labor for pre-treatment and cleaning systems based on the quantity of
equipment specified. It assumes 12 hours of labor per cartridge filter per year. For simplex
basket screens (up to 400 gpm), it assumes 30 minutes of labor per cleaning cycle per screen, to
clean the strainer basket. For automatic self-cleaning prescreens, it assumes 12 hours of labor per
screen per year. Users can change any of these unit labor hours on the O&M assumptions sheet.
The labor associated with cleaning pumps is computed using the approach discussed in
Appendix E.

For other membrane process labor, the model uses a table of estimates provided by a membrane
manufacturer, ranging from 4 hours per week for a 30,000 gal/day system to 200 hours per week
for a 75 MGD system. The table is located on the lookup tables sheet (see Section 3.16).

63
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The sheet calculates maintenance material costs as a percentage of pre-installation capital costs.
For materials for pre-treatment and cleaning, the sheet uses the same percentage as for influent
and booster pumps (1 percent). For the membrane process, it also uses 1 percent of pre-
installation capital costs. Users can change either of these percentages on the O&M assumptions
sheet. Replacement requirements for cartridge filters and membrane elements are calculated
explicitly using the respective useful life assumptions specified on the O&M assumptions sheet
(5 months for cartridge filters and 4 or 5 years for membrane elements). For these components,
the model looks up a quantity discount, based on typical discounts offered by membrane element
distributors. The discount is tabulated on the lookup tables sheet (see Section 3.16).

Chemical usage also is calculated explicitly using the appropriate dosages and the average inflow
(for pre-treatment chemicals) or the cleaning frequency (for cleaning chemicals).

3.14 Indirect Sheet

As stated in Section 2.4, indirect capital costs are costs that are not directly related to the
treatment technology used or the amount or quality of the finished water, but that are associated
with the construction and installation of a treatment technology and water intake structures. The
indirect sheet derives capital costs for the following components of indirect costs:

• Construction management and general contractor overhead

• Standby power
• Geotechnical
• Sitework
• Yard piping.

Appendix D contains detailed information on the derivation of these and other indirect costs.
This sheet also contains calculations to estimate permit costs.

3.15 Output Sheet

The output sheet contains the list of components identified for RO/NF treatment based on the
WBS approach. For each component, the output sheet provides information on size (e.g., tank
capacity or pipe diameter) and quantity, as well as estimated capital cost and estimated useful
life. The output sheet also contains cost estimates for indirect costs (e.g., mobilization and
demobilization, site work and yard piping), add-on costs (for permitting, pilot testing and land)
and O&M costs. These estimates are described generally in Section 2.4 and in more detail in
Appendix D (indirect costs) and Appendix E (O&M costs). Finally, the output sheet combines
the total capital cost, system useful life and annual O&M cost to estimate total annualized cost,
as discussed in Section 2.4. Sections 2.1 and 2.3 provide further details about the output sheet.

As discussed in Section 2.4, the output sheet shows model results to the nearest $1, but this
precision is not meant to imply that the results are accurate to $1. The RO/NF model underwent
peer review in 2007. The majority of peer reviewers who evaluated the model expressed the
opinion that resulting cost estimates would be in the range of budget estimates (+30 to -15
percent). The RO/NF model has since undergone substantial revision in response to the peer
review comments. Users are encouraged to review all documentation, modify inputs and

64
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

assumptions as appropriate to their specific purpose, and form their own informed opinions about
the accuracy and suitability of the results shown on the output sheet.

3.16 Ancillary Model Components

The RO/NF model contains six ancillary sheets: unit conversion factors, membrane
specifications, feed water specifications, concentrate scaling recommendations, cost equations
and lookup tables.

The conversion factors are used to convert input units to metric units for the train design process
and to convert the design results back to input units.

The feed water specifications include the name and water analysis for each feed water. The water
names are displayed in the feed water dropdown on the input sheet. The feed water analysis
includes calcium, magnesium, sodium, potassium, bicarbonate, chlorine, sulfate and fluorine ion
concentrations in parts per million. The silica concentration, pH and silt density index are also
included in the water analysis. The specification table displays the TDS of each water, as the sum
of the listed ions.

The membrane specifications include the manufacturer and name of each available membrane
element, as well as information that the train design process uses to simulate the membrane’s
performance. The various specifications are discussed in Section 3.7. The membrane
specifications in the model were obtained from a study of specification sheets for various
elements and from a study of the how various manufacturer RO design programs model
membrane elements. It is possible for users to add a custom element by pressing the button
labeled “Add New Element” and filling in the required membrane parameters. The bottom row
of the table contains the element names that appear in the input dropdown. It will likely be useful
to copy the parameters from an element that already appears in the model and modify them as
needed.

The concentrate scaling sheet examines the LSI and other saturation fractions. It determines
whether any saturation is great enough to require the use of antiscalant, or whether saturation is
so great that an antiscalant is likely to be ineffective. The input sheet uses its results to display
guidance and warnings. Otherwise, it does not affect the engineering design.

The cost equations sheet uses the component-level cost curve equations from the WBS cost
database to generate unit costs on an item-by-item basis.

The lookup tables sheet contains information used to calculate purchase discounts for pumps and
filters. It also contains information such as charge and molecular weight for the ions in the solute
and a table used to determine the ratio of pressure vessels from one stage of a train to the next.

3.17 List of Abbreviations and Symbols in this Chapter

BaSO4 barium sulfate
CaCO3 calcium carbonate
CaF2 calcium fluoride
CaSO4 calcium sulfate
DBP disinfection byproduct

65
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

EPA U.S. Environmental Protection Agency

gal/day gallons per day
gfd gallons per square foot per day
gpm gallons per minute
HVAC heating, ventilation and air conditioning
lbs/gal pounds per gallon
LSI Langelier saturation index
MGD million gallons per day
mg/L milligrams per liter
NF nanofiltration
O&M operating and maintenance
POTW publicly owned treatment works
psi pounds per square inch
RO reverse osmosis
SDI silt density index
SiO2 silica
SrSO4 strontium sulfate
TDS total dissolved solids
TFC thin-film composite
TFSP Total Flux and Scaling Program
WBS work breakdown structure

3.18 References
AWWA, 1990. American Water Works Association. Water Quality and Treatment: A Handbook
of Community Water Supplies, 4th edition. New York: McGraw-Hill.

AWWA, 1999. American Water Works Association. Water Quality and Treatment: A Handbook
of Community Water Supplies, 5th edition. New York: McGraw-Hill.

AWWA/ASCE, 2005. American Water Works Association and American Society of Civil
Engineers. Water Treatment Plant Design, 4th edition. New York: McGraw-Hill.

AWWARF, 1995. AWWA Research Foundation. Application of Membrane Filtration

Techniques for Compliance with the Surface Water and Groundwater Treatment Rules. October
1995.

AWWARF, 1996. AWWA Research Foundation. Water Treatment Membrane Processes.

Burbano et al., 2006. Burbano, Arturo, Adham, Samer (MWH) and Pearce, William (City of San
Diego). “Current Trends on Membrane-Based Desalination: Findings from a Worldwide Survey
of Full-Scale RO/NF Plants.” Included in the Proceedings of the 2006 AWWA Annual
Conference and Exposition, June 11-15, 2006, San Antonio, Texas.

Cole-Parmer, 2006. Cole-Parmer Catalog, page 673. Accessed online at

<http://www.coleparmer.com/catalog/vcatalog_welcome.asp?from=cwelcom>, June 2006.

66
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Dow, 2005. Dow Liquid Separations FILMTEC Reverse Osmosis Membranes Technical Manual,
July 2005, The Dow Chemical Company, accessed 2006 at
<http://www.dow.com/liquidseps/lit/techman.htm>.

Franks et al, 2005. Franks, Rich N., Bartels, Craig and Wilf, Mark (Hydranautics, Inc.). “The
Effect of Feed Salinity on RO Membrane Salt Passage.” Included in the Proceedings of the 2005
AWWA Membrane Technology Conference, March 6-9, 2005, Phoenix, Arizona.

Freeman and Randtke, 2006. Freeman, Scott (Black and Veatch) and Randtke, Steve (University
of Kansas). Presentation at the workshop “RO and NF Design – Computer Modeling,” at the
2006 AWWA Annual Conference and Exposition, June 11, 2006, San Antonio, Texas.

Jacangelo, J., S. Chellam and R.R. Trussell. 1998. “The Membrane Treatment.” Civil
Engineering.

Mickley, 2006. Membrane Concentrate Disposal: Practices and Regulation. Mickley and
Associates for U.S. Bureau of Reclamation, Desalination and Water Purification Research and
Development Program Report No. 123 (Second Edition), April 2006

Small Business Advocacy Review Panel, 1999. Report of the Small Business Advocacy Review
Panel on EPA’s Planned Proposal of the National Primary Drinking Water Regulation for
Arsenic.

U.S. EPA. 2004a. Estimating the Value of Land for Use in Drinking Water Cost Models. Report
prepared by Science Applications International Corporation. Washington, D.C.: U.S.
Environmental Protection Agency, Office of Groundwater and Drinking Water.

U.S. EPA. 2004b. Evaluating Permit Requirements for Inclusion in Drinking Water Cost
Models. Report prepared by Science Applications International Corporation. Washington, D.C.:
U.S. Environmental Protection Agency, Office of Groundwater and Drinking Water.

U.S. EPA. 2011. Pilot and Bench Study Costs for Drinking Water Rules. Report prepared by
Science Applications International Corporation. Washington, D.C.: U.S. Environmental
Protection Agency, Office of Groundwater and Drinking Water.

Watson, 2002. Watson, Ian (AEPI-RosTek). Review of critical design assumptions, October
2002.

Wolfe, 2004. Wolfe, Thomas (WaterEye Corp.). Personal communication, November, 2004.

Wolfe, 2006. Wolfe, Thomas (WaterEye Corp.). Presentation at the workshop “RO and NF
Design – Computer Modeling,” at the 2006 AWWA Annual Conference and Exposition, June
11, 2006, San Antonio, Texas.

67
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Appendix A. Valves, Instrumentation and System Controls

A.1 Valves in the WBS Models
There are many types of valves used to control water and chemical flow rates, pressure and
direction in a water treatment plant. Valves can be distinguished by function, mode of operation,
materials of construction, size (i.e., diameter), design or shape and connection method. For
purposes of estimating valve costs, the most important of these distinctions are function, mode of
operation, size and materials of construction. Therefore, the work breakdown structure (WBS)
models group valves according to these four distinctions. The WBS models identify valve size
explicitly, using the same methodology used to size pipes. The WBS models also explicitly
identify materials of construction. The output sheet of each model includes line item costs for
valves of each material (plastic, stainless steel and cast iron), so that the user can observe
variations in cost among materials.

To distinguish by function and mode of operation, the WBS models use a generic nomenclature.
The WBS models identify valves as one of the following:

• Check valves
• Manual valves
• Motor/air-operated valves.

Check valves are those that serve the function of backflow prevention. They generally do not
vary significantly in mode of operation or design/shape.

The other two categories of valves serve the function of flow control and are distinguished by
their mode of operation (i.e., whether they are manual or automated). An example of a valve that
must be a manual valve is an emergency shut-off valve that, in an extreme event such as
complete power failure, can be shut off by an operator. Manual valves can vary in design
according to their specific opening/closing method (e.g., hand wheel or chain). Automated valves
(identified in the WBS models as motor/air operated) can be motor-operated valves, air-operated
valves or solenoid valves. Solenoid valves are electrically operated on/off control valves. Motor-
operated valves open and close more slowly than solenoid valves. This action reduces likelihood
of a water hammer. While the different opening/closing methods for manual and automated
valves have various advantages and disadvantages, cost differences among designs are relatively
small and the WBS unit costs do not distinguish between them at this level of detail. The key
cost difference is whether the valves are automated or manual, because of the cost of the motor,
air actuator or solenoid.

A.2 Instrumentation for Process Measurements

Each of the models includes the cost of various instruments that perform process measurements.
Most of these measurement devices are categorized into the following groups:

• Hydraulic measurement instruments and control devices. Hydraulic measurement

instruments include: flow meters, pressure gauges, head loss sensors and water level
meters/alarms. Hydraulic control devices include: pump control, motor control and valve
control.

68
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• Water quality measurement and control devices. These include water quality parameter
measurement devices, such as pH meters, oxidation-reduction potential (ORP) sensors,
temperature meters, turbidity meters and sampling devices and ports.

The WBS models determine instrumentation requirements for each technology based on review
of the schematic flow diagram for the appropriate technology, along with certain general
assumptions that are applied to all of the technologies. Exhibit A-1 documents the general
assumptions about instrumentation that are applied in the WBS models. Slightly different
assumptions hold when a model is intended as an add-on to an existing process (e.g., acid feed)
rather than a complete process (e.g., anion exchange).

Exhibit A-1. General Design Assumptions for Instrumentation

Instrument Type Assumption
Chlorine analyzers For chlorine and hypochlorite disinfection, 1 per treatment train to monitor residual
Conductivity meters Varies by technology
Dissolved oxygen analyzers Varies by technology
Drive controllers 1 per each pump (including booster, backwash and chemical metering pumps) or
other motorized item of equipment (e.g., mixers, blowers) in fully automated systems
Electric enclosures Only for technologies with significant electric-powered equipment outside a building
structure
Flow meters 1 for the influent or effluent line and 1 for backwash discharge. Some technologies
also include flow meters on process lines.
Head loss sensors Continuous level sensors. 1 per process vessel for technologies with pressure
vessels. Some technologies omit head loss sensors for systems with design flows
less than 1 million gallons per day.
High/low alarms 1 per backwash tank and 1 per chemical storage tank
Level switch/alarms 1 per process basin; 2 per contact tank for chemical disinfection technologies.
Technologies with chemical cleaning use 1 per chemical tank in the cleaning system.
ORP sensors Varies by technology
Particle meters Varies by technology
pH meters 1 each for the influent and effluent lines for systems with pH adjustment, plus others
on a technology-specific basis
Pressure transducers Included in the cost of flow meter assemblies for venturi and orifice plate meters
Sampling ports 1 per process vessel, plus 1 each for the influent line, effluent line and discharge side
of the backwash line for complete process models. Others are included on a
technology-specific basis.
Total dissolved solids monitors Varies by technology
Temperature meters Varies by technology; often 1 for the influent and/or effluent lines, except for add-on
models. Some technologies omit temperature meters for systems with design flows
less than 1 million gallons per day.
Total organic carbon analyzers Varies by technology
Turbidity meters Varies by technology

Several types of flow meters can appear in the model output: propeller, venturi, orifice plate and
magnetic flow meters. In general, the choice of meter depends on the cost level and design flow
of the system, although some technologies require particular types of flow meters for specific
purposes. For smaller and/or low-cost systems, the preference order in the models will have
propeller flow meters as a first choice; for intermediate systems, venturi flow meters top the
preference order; and for larger and/or high-cost systems, the top preference is magnetic flow

69
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

meters. In all cases, the component buildup will display the price for all available types of flow
meters at a given size, so that a user can assess the cost impact of different types.

The critical design assumptions sheet of each model incorporates these general assumptions.
Therefore, the user can adjust instrumentation assumptions on a technology-specific basis.

Individual technologies—in particular, aeration technologies and chlorine and hypochlorite

disinfection—have additional or differing instrumentation requirements.

A.3 Control Systems

Automated control systems comprise the hardware and software used to monitor and control a
treatment process. There are two general types of systems: programmable logic controls (PLCs)
and/or remote telemetry units (RTUs). PLCs are stand-alone microprocessor-based control
systems that can be programmed to monitor and control process equipment. RTUs were
originally developed to communicate with systems from remote, outdoor locations. Newer RTU
models can provide full equipment control through remote operator interface (AWWA, 2001).
Because the WBS cost models (except for the nontreatment model) pertain to centralized
treatment facilities, the assumptions reflect the control of all system components using a PLC
system; RTUs are generally more appropriate for remote communications.

PLC hardware consists of a rack-mounted system with plug-in slots for the input and output
(I/O) modules, which provide connections for the instruments and equipment, and one or more
central processing unit (CPU) modules, which process the monitoring data inputs and control
command outputs. The PLC equipment requires a power supply unit to operate the PLC data and
command processing functions. In addition, an uninterruptible power supply (UPS) will protect
the PLC system from undesired features such as outages and surges that can adversely affect the
performance of the PLC unit. A system operator can monitor and operate with the PLC using
either a computer or an operator interface unit, which is a panel mounted on the PLC enclosure.
These units can be as simple as 2-line light-emitting diode text panels or as advanced as full
color touch panels. The WBS models have default assumptions that PLC systems for smaller
drinking water systems will be operated using an advanced, fully-functional operator interface
unit after the control system installer has programmed the PLC. Larger systems will include an
operator interface unit with more limited functionality and use at least one computer workstation
with PLC programming software and printers to accomplish more advanced control functions
from a central location. Large systems also include plant intelligence software to assist operation
of the extensive control system.

The PLC system design in the WBS models depends on the design of the treatment system,
which dictates the total number and type of I/O connections. The PLC system receives input
signals from and transmits output signals to ports on instruments and equipment controllers. The
I/O signals may be discrete or analog, depending on the type of equipment generating or
receiving the signals. Discrete signals indicate which of two conditions apply such as whether a
switch is on or off. Analog signals indicate a value along a predefined range such as temperature
or rate of flow. Exhibit A-2 identifies the assumptions used in the WBS models to determine the
total number of I/O connections required for the PLC system.

70
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit A-2. I/O Port Requirements for Instrumentation and Control

Instrument Type Inputs to and Outputs from PLC System
Alarm (level switch/alarm, high/low alarm or low alarm) 1 input and 1 output—discrete
Chlorine analyzer 1 input—analog
Conductivity meter 1 input—analog
Dissolved oxygen analyzer 1 input—analog
Drive controller 3 inputs (1 for the auto switch position, 1 for the run status signal
and 1 for overload or fault signal) and 1 output—discrete
Flow meter 1 input—analog
Venturi and orifice plate meters also include inputs and outputs
for the associated pressure transducer (below)
Head loss sensor 1 input—discrete
Motor/air-operated valve 1 input and 1 output—analog
ORP sensor 1 input—analog
Particle meter 1 input—analog
pH meter 1 input—analog
Pressure transducer 1 input—analog
Sampling port 1 input—discrete
Total dissolved solids monitor 1 input—discrete
Temperature meter 1 input—analog
Total organic carbon analyzer 1 input—analog
Turbidity meter 1 input—analog

The degree of automated control at a treatment facility can range from none to a fully automated
control system that can monitor and control the hydraulic regime at the plant, the chemicals
addition system, the power system and the communication system. To reflect potential ranges in
treatment costs, the WBS models can provide equipment and operator labor cost outputs for
three degrees of control:

• Fully automated control with safety overrides

• Semi-automated control where instruments provide data and information to the control
station, but operators manually activate valves and mechanical equipment (e.g., this
option removes outputs from the PLC system and removes automated drive controllers
from mechanical equipment)
• Fully manual control where operators collect data directly from the instruments and
manually activate valves and mechanical equipment.

Users can select among these three control schemes using the system automation input in each
WBS model (see Section 2.3). Exhibit A-3 shows the general design assumptions about control
equipment used for each control scheme in the WBS models. The paragraphs below provide
additional information regarding the equipment components and calculations.

71
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit A-3. General Design Assumptions for System Controls

Small Medium Large
System System System
Item of Control Equipment (<1 MGD) (1–10 MGD) (>10 MGD) WBS Assumption
PLC Rack/Power Supply A,S A,S A,S 1 base and expansion bases
as needed for I/O (see text)
CPU A,S A,S A,S 2 per system1
I/O Discrete Output Module A A A 1 for every 32 outputs2
I/O Discrete Input Module A,S A,S A,S 1 for every 32 inputs2
I/O Combination Analog Module A,S A,S A,S 1 for every 12 inputs (for A
and S) and outputs (for A
only)3
Ethernet Module A,S A,S A,S 2 per system1
Base Expansion Module A,S A,S A,S 1 per expansion base
Base Expansion Controller Module A,S A,S A,S 1 per expansion base
UPS A,S A,S A,S 1 per system
Operator Interface Unit – limited NA A,S A,S 2 per system1 (see text)
functionality
Operator Interface Unit – advanced, A,S NA NA 2 per system1 (see text)
fully functional
Computer Workstations NA A,S A,S 1 per operator
Laser Jet Printer NA A,S A,S 1 per 4 workstations
Dot Matrix Printer NA A,S A,S 1 per 4 workstations
PLC Programming Software NA A,S A,S 1 per workstation
Operator Interface Software A,S NA NA 1 per system
PLC Data Collection Software NA A,S A,S 1 per workstation
Plant Intelligence Software NA A,S A,S 1 per workstation
A = included in a fully automated system; S = included in a semi-automated system; NA = not applicable for this design size
Notes: Fully manual systems do not include system controls
1. Includes one to provide redundancy
2. Discrete input and output modules can have fewer I/O connections, but price differences are small. To keep the equipment
requirement calculation tractable, the WBS models use a 32-connection module, which will slightly overstate cost when fewer
connection points are needed on the last module.
3. A combination module accommodates 8 inputs and 4 outputs. This 2-to-1 ratio is generally consistent with the ratio of analog
inputs-to-outputs in the WBS models for a fully automated system.

The primary PLC system is a rack and power supply (i.e., a “base”) with nine slots for control
modules. 9 The CPU module requires one slot. An ethernet module necessary for PLC
programming requires a second slot, leaving seven for I/O modules. If additional I/O slots are
needed to accommodate instruments and equipment, then up to four additional expansion bases
can be added, giving the single CPU the capacity to run up to 8,192 I/O connections. Each
expansion base has nine module slots and is linked to the CPU module on the primary base.

The total number of PLC racks and power supplies include the primary rack and any expansion
racks. The calculation for the total number of racks must take into account the module slots that
will be occupied by all types of modules including the CPU module, the ethernet module and
expansion base controller modules. Each expansion rack requires a base expansion controller
module, which occupies one of the module slots on the expansion rack, leaving eight slots for

9
Bases with fewer slots are also available, but cost differences across base sizes are small. To keep the equipment
requirement calculation tractable, the WBS models use a 9-slot base, which will slightly overstate cost when fewer
slots are needed.

72
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

I/O modules. Each expansion rack also requires a base expansion module, which is attached to
the outside of the rack and, therefore, does not require a module slot. The following calculations
illustrate how the WBS models calculate total PLC racks:

IF (plc_cpu + plc_ethernet + plc_discrete_input + plc_discrete_output +

plc_combination_analog) ≤ 9
THEN plc_rack = 1, plc_base_expansion = 0, plc_ base_expansion_controller = 0

IF (plc_cpu + plc_ethernet + plc_discrete_input + plc_discrete_output +

plc_combination_analog) > 9 AND ≤ 17
THEN plc_rack = 2, plc_base_expansion = 1, plc_ base_expansion_controller = 1

IF (plc_cpu + plc_ethernet + plc_discrete_input + plc_discrete_output +

plc_combination_analog) > 17 AND ≤ 25
THEN plc_rack = 3, plc_base_expansion = 2, plc_ base_expansion_controller = 2

IF (plc_cpu + plc_ethernet + plc_discrete_input + plc_discrete_output +

plc_combination_analog) > 25 AND ≤ 33
THEN plc_rack = 4, plc_base_expansion = 3, plc_ base_expansion_controller = 3

IF (plc_cpu + plc_ethernet + plc_discrete_input + plc_discrete_output +

plc_combination_analog) > 36 AND ≤ 41
THEN plc_rack = 5, plc_base_expansion = 4, plc_ base_expansion_controller = 4

A.4 List of Abbreviations and Symbols in this Appendix

CPU central processing unit
I/O input and output
ORP oxidation-reduction potential
PLCs programmable logic controls
RTUs remote telemetry units
UPS uninterruptible power supply
WBS work breakdown structure

A.5 References
American Water Works Association (AWWA). 2001. Instrumentation and Control, Manual of
Water Supply Practices—M2. Third Edition. Denver, Colorado: AWWA.

73
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Appendix B. Building Construction Costs

B.1 Introduction
The work breakdown structure (WBS) cost database incorporates building costs from three
sources: RSMeans 2009 Square Foot Costs (RSMeans, 2008), Saylor 2009 Commercial Square
Foot Building Costs (Saylor, 2009) and the Craftsman 2009 National Building Cost Estimator
(NCBE) software model (described in Craftsman, 2008). Each of these sources enables a user to
create a cost estimate by combining costs for different elements of a building—for example, the
foundation, exterior walls or light fixtures.

For each source, the WBS cost database includes three sets of options to represent low, medium
and high building design and material qualities. Section B.2 provides descriptions of the relevant
options for each source within these categories, as well as the options selected for each of the
three building types used in the WBS models. The WBS models cost heating and cooling
systems as individual capital cost line items separate from the building construction costs.
Therefore, the building costs discussed here exclude heating and cooling systems.

For each of the three types of building, EPA used the three sources described here to develop
cost buildups for 24 building sizes ranging from 500 to 200,000 square feet and tabulated costs
for each of the models. The resulting costs from each model are included in the WBS cost
database. The database escalates these costs from 2008 dollars using the Engineering News-
Record Building Cost Index (ENR, 2013) and averages them following the same procedure as
for other components, as described in Chapter 2. The WBS models use these costs to estimate
costs per square foot for buildings larger than 500 square feet (ft2).

EPA also developed a fourth building type that applies only to structures smaller than 500 ft2—
essentially a shed with steel walls and a roof. This additional building type allows the WBS
models to use, for very small systems, building costs that reflect very inexpensive building
construction methods and materials. For this type of building, EPA used the Craftsman NCBE
model to estimate costs for a low-profile steel building. However, the WBS models do not use
this building type for chlorine storage buildings because chlorine gas use necessitates a non-
corrodible building material and special ventilation requirements. Thus, for chlorine storage
buildings smaller than 500 ft2, the WBS models use the same unit costs as for larger buildings.

B.2 Buildup Options and Building Quality Selections

EPA developed building cost estimates using comparable assumptions across data sources: the
Craftsman NBCE model, building costs from RSMeans 2009 Square Foot Costs and Saylor 2009
Commercial Square Foot Building Costs. Each source provides unit costs for different building
types and construction qualities.

The Craftsman NBCE model is a software model that generates building cost estimates based on
user input (i.e., building size and quality of building features and fixtures). Given the variation in
unit costs for components by size, it appears to function as a parametric model. The costs in the
NBCE model are based on data obtained from U.S. government building cost surveys.

74
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The RSMeans and Saylor manuals contain unit costs, usually in dollars per square foot, for
various building components (e.g., exterior walls, floor structure or roof structure). The costs are
based on data obtained from the construction industry and independent research of construction
costs. By combining unit costs across components, one can build up a total building unit cost.
The approach is essentially a WBS cost approach where most components are priced on the basis
of building area, with little or no variation in the cost per unit area as building size increases. For
example, the RSMeans unit cost for a foundation slab varies with the thickness of the slab (EPA
chose thicker slabs for higher quality buildings), but not with the building size. Notable
exceptions are the cost of exterior walls and roof structures. Exterior wall cost in dollars per ft2
declines as building size increases because the ratio of exterior wall linear footage to square
footage declines. For roof structures, EPA chose roof spans based on the length of a side of the
building (assumed square). For building side lengths greater than 70 feet, EPA included support
columns to give a maximum roof span of 70 feet. Larger buildings, therefore, may have
somewhat more expensive superstructures on a per-square-foot basis, since they may have a
wider roof span or support columns.

EPA chose inputs to the NBCE model and chose components from the RSMeans and Saylor
manuals to reflect the different levels of building quality used in the WBS models (high,
medium, low and very small low quality).

Based on the NBCE industrial building quality classifications, EPA determined that the NBCE
Class 1&2 (best/good quality), Class 3 (average quality) and Class 4 (low quality) reflected WBS
high, medium and low quality buildings, respectively. EPA used the NBCE low-profile steel
building for very small low quality buildings.

The RSMeans and Saylor manuals do not contain building types that are closely comparable to
the very small low quality building. Therefore, there are no RSMeans or Saylor costs for this
type of structure. RSMeans and Saylor building cost estimates were “built” by selecting specific
building elements of differing quality for each type of building from the assemblies sections of
their respective manuals.

From each source, EPA obtained cost estimates for the following building areas in square feet:
500; 1,000; 2,000; 3,000; 4,000; 5,000; 7,500; 8,000; 10,000; 12,000; 15,000; 18,000; 20,000;
24,000; 25,000; 30,000; 36,000; 42,000; 48,000; 50,000; 54,000; 60,000; 100,000 and 200,000.
The resulting costs do not include costs for site improvements (e.g., land, landscaping, parking
and utilities), permits, furnishings and production equipment, homeland security responses or
contingency allowance.

The RSMeans and Saylor costs include installation costs as well as overhead and profit for the
contractors installing the building components, but do not include architectural fees or general
contractor markup for general conditions, overhead and profit (RSMeans, 2008; Saylor, 2009).
According to Craftsman (Ogershok, 2009), the NBCE model’s costs do not include installing
contractor markup directly, but do include a markup of 30 percent for the general contractor,
which they assume to also cover the installing contractor’s markup. Since the Craftsman costs
were generally lower than those from Means or Saylor and since the installing contractor’s
markup in Means and Saylor is usually 30 percent or more, EPA assumed that the 30 percent
markup in the Craftsman costs was passed along directly to the installing contractor and further

75
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

markup would be required for the general contractor. Architectural fees and the general
contractor’s markup are included in the WBS model indirect cost output, as described in
Appendix D.

Each source has a different set of options. They can be grouped into six categories:

• Substructure
• Superstructure
• Exterior closure
• Interior finish
• Mechanical services, excluding heating and cooling
• Electrical services.

B.2.1 Substructure
Building substructure was selected using application scenarios for each of the three quality
options. For low quality buildings, EPA assumed an average industrial use scenario. For medium
quality buildings, EPA assumed a heavy industrial use scenario. For higher quality buildings,
EPA assumed a heavy industrial with live loads use scenario. EPA assumed light foot traffic for
the very small (less than 500 ft2) buildings (other than those used to store chlorine gas).

Exhibit B-1 shows the detailed choices that EPA made for each of the three sources.

Exhibit B-1. Substructure Selections for NBCE, RSMeans and Saylor

Building Medium Quality Very Small Lower
Lower Quality Building Higher Quality Building
Variable Building Quality Building
Foundation: reinforced
Foundation: continuous Foundation: continuous Foundations as required
concrete pads under
reinforced concrete. reinforced concrete. for normal soil conditions;
Craftsman pilasters.
Floor: 6” rock base, 5” Floor: 6” rock base, 6” a 4” concrete floor with
NBCE Floor: 6” rock base, 4”
concrete with reinforcing concrete with reinforcing reinforcing mesh and a 2”
concrete with reinforcing
mesh or bars. mesh or bars. sand fill.
mesh.
Foundation: poured Foundation: poured Foundation: poured
concrete; strip and spread concrete; strip and spread concrete; strip and spread
footings. footings. footings.
Slab: 4” reinforced, Slab: 5” reinforced, heavy Slab: 6” reinforced, heavy
industrial concrete with industrial concrete with industrial concrete with
RSMeans not applicable
vapor barrier and granular vapor barrier and granular vapor barrier and granular
base. Site preparation for base. Site preparation for base. Site preparation for
slab and trench for slab and trench for slab and trench for
foundation wall and foundation wall and foundation wall and
footing. 4’ foundation wall. footing. 4’ foundation wall. footing. 4’ foundation wall.
Foundation: concrete strip Foundation: concrete strip Foundation: concrete strip
and spread footings, 4' and spread footings, 4' and spread footings, 4'
foundation wall. foundation wall. foundation wall.
Saylor Slab on grade: reinforced Slab on grade: reinforced Slab on grade: reinforced not applicable
concrete, vapor barrier, 4" concrete, vapor barrier, 5" concrete, vapor barrier, 6"
thick, on 4' sand or gravel thick, on 4' sand or gravel thick, on 4' sand or gravel
base. base. base.
‘ = feet; “ = inches

76
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

B.2.2 Superstructure
EPA assumed the same quality of superstructure for each of the three quality options—metal
deck and open web steel joists, supported by columns and exterior walls. However, the
superstructure support column spans range up to 70 feet, depending upon building size. To
establish the column span, EPA computed the length of a building side, assuming the building to
be square. For buildings with side lengths larger than 70 feet, EPA included support columns in a
square grid to provide a roof span of 70 feet or less, assuming that the roof would also be
supported on the exterior walls. For instance, a 10,000 ft2 building (100 feet on a side) would
have one support column in the center, with a 50 foot roof span. A 30,000 ft2 building (173 feet
on a side) would have four support columns at 58 foot intervals. Since the sources included roof
spans in increments of 10 feet, EPA rounded up to a 60 foot roof span for this building. EPA
used a steel building quality superstructure for the very small (less than 500 ft2) buildings (other
than those used to store chlorine gas).

Exhibit B-2 displays the superstructure options that EPA selected for each source.

Exhibit B-2. Superstructure Selections for NBCE, RSMeans and Saylor

Very Small
Building Lower
Lower Quality Building Medium Quality Building Higher Quality Building
Variable Quality
Building
Roof structure: glu-lams wood or Roof structure: glu-lams Roof structure: glu-lams
steel trusses on steel intermediate wood or steel trusses on wood or steel trusses on Steel roof
columns, short span. Roof cover: steel intermediate columns, steel intermediate columns, purlins 4½
panelized roof system, ½” plywood short span. Roof cover: span exceeds 70’. to 5½ feet
Craftsman sheathing, 4-ply built-up roof. 10 ft2 panelized roof system, ½” Roof cover: panelized roof on centers,
NBCE of skylight per 2,500 ft2 of floor plywood sheathing, 4-ply system, ½” plywood 26-gauge
area (1-2’x 4’skylight 40’ to 50’ built-up roof. 24 ft2 of skylight sheathing, 4-ply built-up galvanized
o.c.). per 2,500 ft2 of floor area (1- roof. 32 ft2 of skylight per steel on
4’x 6’ skylight 40’ to 50’ o.c.). 2,500 ft2 of floor area (1-4’x roof
8’ skylight 40’ to 50’ o.c.).
Roof: 1.5” galvanized metal deck,
open web steel joists, joist girders,
on columns and walls; total load =
60-65 lbs/ft2. Column spacing
chosen to give a maximum span of
not
RSMeans 70’, with the building assumed Same as lower quality Same as lower quality
applicable
square. Steel columns.
Roof cover: Built-up tar and gravel
roof covering with flashing,
perlite/EPS composite insulation.
Roof hatches with curb.
Roof: metal deck, open web steel
joists, on columns and walls. Wide
flange steel columns, steel beams
and girders. Column spacing not
Saylor Same as lower quality Same as lower quality
chosen to give a maximum span of applicable
70’, with the building assumed
square.
Roof cover: built-up tar and gravel.
‘ = feet; “ = inches; EPS = expanded polystyrene; o.c. = on center

77
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

B.2.3 Exterior Closure

EPA used different building exterior qualities to estimate unit costs that vary by exterior
material. EPA selected reinforced concrete block exteriors for the lower quality buildings,
reinforced tilt-up concrete panel exteriors for the medium quality buildings 10 and brick-faced,
reinforced cavity/composition wall exteriors for the higher quality buildings. EPA used
corrugated metal exteriors for the very small lower quality structures (smaller than 500 ft2).

A cavity wall (e.g., masonry) is a wall in which the inner and outer wythes are separated by an
air space, but tied together with wires or metal stays. A composition wall is a wall combining
different materials to work as a single unit. A tilt-up wall is a method of concrete construction in
which wall sections are cast horizontally at a location adjacent to their eventual position and
tilted into place after removal of forms.

Exhibit B-3 shows the exterior closure options that EPA selected for each model.

Exhibit B-3. Exterior Closure Selections for NBCE, RSMeans and Saylor
Building Medium Quality Very Small Lower
Lower Quality Building Higher Quality Building
Variable Building Quality Building
Steel frames/bents set 20’
8” reinforced concrete
to 24’ on centers, steel
8” reinforced concrete 8” reinforced concrete block or brick with
wall girts 3½’ to 4½’ on
Craftsman block or brick, unpainted. block or brick, unpainted. pilasters 20’ on centers,
centers, post and beam
NBCE (Same for both lower and (Same for both lower and painted sides and rear
type end wall frames, 26-
medium quality.) medium quality.) exterior, front wall brick
gauge galvanized steel
veneer
on ends and sides
Brick face composite wall-
Concrete block, Tilt-up concrete panels, double wythe: utility brick,
RSMeans reinforced, regular weight, broom finish, 5½” thick, concrete block backup not applicable
hollow, 4x8x16’, 2,000 psi 3,000 psi masonry, 8” thick, perlite
core fill.
Concrete block, 4x8x16’, Tilt-up concrete panel, 6" Brick cavity wall,
Saylor not applicable
reinforced thick, no pilasters. reinforced, 10" thick.
‘ = feet; “ = inches; psi = pounds per square inch

B.2.4 Interior Finish

Choices of interior finish reflect the quality and duty of the interior construction materials such
as floor coverings, wall coverings and ceilings. EPA selected functional, minimally attractive
interior finishes for the lower quality buildings and more functional and attractive interiors for
medium and higher quality buildings. EPA also selected functional, unattractive interior finishes
for the very small (less than 500 ft2) buildings (other than those used to store chlorine gas).

Exhibit B-4 shows the interior finish options that EPA selected for each source.

10
Tilt-up concrete panel exteriors were selected in the RSMeans and Saylor cost estimation buildups. Tilt-up
concrete panels were not an exterior option in the Craftsman NBCE cost estimation model; therefore, reinforced
concrete block exterior was selected in the Craftsman NBCE cost estimation model for medium quality buildings.

78
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit B-4. Interior Finish Selections for NBCE, RSMeans and Saylor
Building Medium Quality Very Small Lower
Lower Quality Building Higher Quality Building
Variable Building Quality Building
Concrete floors. Rest
Concrete floors. Rest Concrete floors. Rest
rooms: enameled gypsum Minimal quality, minimal
Craftsman rooms: unfinished rooms: painted gypsum
wallboard partitions, 3 duty, functional,
NBCE wallboard partitions and 2 wallboard partitions and 2
good fixtures, vinyl unattractive
low cost fixtures. average fixtures.
asbestos tile floors.
One minimal quality 2- One high quality 3-fixture
One minimal quality 2-
fixture restroom per 5,000 restroom per 5,000 ft2
fixture restroom per 5,000
ft2 building area. Painted building area. Acrylic
ft2 building area.
walls. Vinyl composition glazed walls. Vinyl
Unpainted walls.
tile floors covering 10 composition tile floors
RSMeans Concrete floors. not applicable
percent of building area. covering 10 percent of
Fiberglass ceiling board
Fiberglass ceiling board building area. Fiberglass
on exposed grid system
on exposed grid system ceiling board on exposed
covering 10 percent of
covering 10 percent of grid system covering 10
building area.
building area. percent of building area.
One restroom per 5,000 One restroom per 5,000
One restroom per 5,000
ft2 building area, with 2 ft2 building area, with 3
ft2 building area, with 2
standard fixtures, baked standard fixtures, baked
economy fixtures, baked
enamel partitions. Painted enamel partitions. Painted
enamel partitions.
walls. Vinyl composition walls. Vinyl composition
Saylor Unpainted walls. not applicable
floor covering 10 percent floor covering 10 percent
Concrete floors. Ceiling:
of building area. Ceiling: of building area. Ceiling:
5/8” gypsum board on
5/8” gypsum board on 5/8” gypsum board on
metal frame, covering 10
metal frame, covering 10 metal frame, covering 10
percent of building area.
percent of building area. percent of building area.
‘ = feet; “ = inches

B.2.5 Mechanical Services

Mechanical services include fire protection, plumbing, heating, ventilation and cooling. The
WBS models cost heating and cooling systems as individual capital cost line items separate from
the building construction costs, so the mechanical services included in the building costs are
limited to fire protection, plumbing and ventilation. EPA assumed no sprinkler systems for the
lower quality buildings and normal hazard wet sprinkler systems for medium and higher quality
buildings. EPA also assumed no sprinkler systems for the very small (less than 500 ft2) buildings
(other than those used to store chlorine gas).

Exhibit B-5 shows the mechanical services options that EPA selected for each source.

79
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit B-5. Mechanical Services Selections NBCE, RSMeans and Saylor

Building Medium Quality Very Small Lower
Lower Quality Building Higher Quality Building
Variable Building Quality Building
Sprinklers. 1 medium Sprinklers. 1 medium
No sprinklers. 1 small rotary vent per 2,500 ft2 of rotary vent per 2,500 ft2 of Minimal quality, minimal
Craftsman
rotary vent per 2,500 ft2 of floor area. (Same for both floor area. (Same for both duty, functional, no
NBCE
floor area. medium and higher medium and higher sprinklers
quality.) quality.)
Gas-fired water heater. Gas-fired water heater.
Wet pipe sprinkler Wet pipe sprinkler
Gas-fired water heater.
RSMeans system. (Same for both system. (Same for both not applicable
No sprinklers.
medium and higher medium and higher
quality.) quality.)
Gas-fired water heater (1 Gas-fired water heater (1
per 5,000 ft2), 50 gallon, per 5,000 ft2), 50 gallon,
Gas-fired water heater (1 100 GPH. Exposed wet 100 GPH. Exposed wet
Saylor per 5,000 ft2), 50 gallon, sprinkler system, normal sprinkler system, normal not applicable
100 GPH. No sprinklers. hazard. (Same for both hazard. (Same for both
medium and higher medium and higher
quality.) quality.)
‘ = feet; “ = inches; GPH = gallons per hour

B.2.6 Electrical Services

EPA included the cost of light fixtures and convenience power, along with associated wiring and
conduits. EPA selected inexpensive lighting fixtures that provide minimal lighting and a minimal
number of wall switches and receptacles for the lower quality buildings and selected increasingly
expensive lighting fixtures that provide bright lighting and an increased number of wall switches
and receptacles for the medium and higher quality buildings. EPA also selected minimal lighting
fixtures for the very small (less than 500 ft2) buildings (other than those used to store chlorine
gas).

EPA did not include electrical feed, switchgear, motor control centers, etc. in building costs.
These costs are likely to vary significantly by technology for buildings of the same size and
quality; for example, a mid-sized reverse osmosis system and a small packaged conventional
filtration system might occupy roughly the same footprint in similar buildings, but the reverse
osmosis system will likely have much greater power requirements. It is therefore not appropriate
to base these costs on the building’s area or quality. These costs are included in the indirect cost
buildup based on a percentage of process cost, as described in Appendix D.

Exhibit B-6 shows the electrical services options that EPA selected for each source.

80
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit B-6. Electrical Services Selections for NBCE, RSMeans and Saylor
Building Medium Quality Very Small Lower
Lower Quality Building Higher Quality Building
Variable Building Quality Building
Lighting: low cost Lighting: low cost single Lighting: 4” single tube Minimal quality, minimal
Craftsman
incandescent fixtures, tube fluorescent fixtures fluorescent fixtures duty, basic wiring and
NBCE
20’x30’ spacing 20’x20’ spacing 10’x12’ spacing minimal lighting fixtures
Lighting: Incandescent Lighting: Fluorescent Lighting: Fluorescent
fixtures recess mounted, fixtures recess mounted fixtures recess mounted
type A: 1 W/ft2, 8 FC. 6 in ceiling: T-12, 40 W in ceiling: T-12, 40 W
lighting fixtures, 1 wall lamps, 2 W/ft2, 40 FC. 10 lamps, 4 W/ft2, 80 FC. 20
RSMeans switch and 2.5 lighting fixtures, 2.5 wall lighting fixtures, 5 wall not applicable
receptacles per 1,000 ft2. switches and 5 switches and 10
1 W miscellaneous receptacles per 1,000 ft2. receptacles per 1,000 ft2.
power. 1.5 W miscellaneous 3 W miscellaneous
power. power.
Lighting: Incandescent Lighting: Fluorescent Lighting: Fluorescent
fixtures, surface mounted, fixtures, recessed, 2 13 W fixtures, recessed, 2 13 W
100 W, commercial bulbs each, 16 per 1,000 bulbs each, 31 per 1,000
grade, 10 per 1,000 ft2 forft2 for 2 W/ft2 total. 2.5 ft2 for 4 W/ft2 total. 5
1 W/ft2 total. 1 commercial grade single- commercial grade single-
commercial grade single- pole switches and 5 pole switches and 10
pole switch and 2.5 commercial-grade duplex commercial-grade duplex
Saylor not applicable
commercial-grade duplex receptacles per 1,000 ft2. receptacles per 1,000 ft2.
receptacles per 1,000 ft .
2 EMT conduit and wire for RGS conduit and wire for
In slab/PVC conduit and 60 A current, length 60 A current, length
wire for 60 A current, assumed equal to assumed equal to
length assumed equal to building perimeter for a building perimeter for a
building perimeter for a square building. square building.
square building.
‘ = feet; “ = inches; A = amp; EMT = electrical metallic tubing; FC = foot candles; PVC = polyvinyl chloride; RGS = rigid
galvanized steel; W = watt

B.3 List of Abbreviations and Symbols in this Appendix

EPA U.S. Environmental Protection Agency
ft2 square feet
NBCE National Building Cost Estimator
WBS work breakdown structure

B.4 References
Craftsman Book Company. 2008. 2009 National Building Cost Manual: 33rd Edition. October.

Engineering News-Record (ENR). 2013. Building and Construction Cost Indexes. Online at
http://enr.construction.com/economics/

Ogershok, Dave, Craftsman Book Company. 2009. Personal communication with Danielle Glitz,
SAIC. 6 March.

RSMeans. 2008. 2009 Square Foot Costs. 30th Annual Edition. Kingston, Massachusetts:
RSMeans Company.

81
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Saylor Publications, Inc. 2009. 2009 Commercial Square Foot Building Costs: 19th Annual
Edition.

82
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Appendix C. Residuals Management Costs

C.1 Introduction
The purpose of this appendix is to outline the approach used to estimate costs for managing the
residuals generated by different drinking water treatment technologies. The work breakdown
structure (WBS) model for each treatment technology includes its own residuals cost estimate.
Each model allows the user to choose from different residual management options that reflect the
methods most likely to be used for the drinking water treatment technology being modeled.
Based on the residuals management option selected, each model identifies the specific
component equipment and operating and maintenance (O&M) requirements and generates costs
using the WBS approach based on engineering design. Costs for residuals management
equipment appear as line items in the model output, as is the case for other WBS elements. The
residuals management design also affects indirect costs, land costs and building costs.

The residuals management options available in each model are specific to the technology being
modeled, driven by the types of residuals generated, their quantity, the frequency of generation
(e.g., intermittent versus continuous) and their characteristics. There are, however, similarities
among groups of technologies that generate similar residuals. Exhibit C-1, below, lists the
technology groups, the residuals generated and the frequency of generation.

The technology-specific chapters of this report identify the residuals management options
available in each model. Because many of the options are similar within (or even across)
technology groups, this appendix describes the methodology and assumptions used for each
option in a single location, rather than repeating the information in each technology chapter. The
residuals management options that may be included in a given model include the following:

• Holding tanks (with or without coagulant addition)

• Direct discharge to surface water
• Discharge to a publicly owned treatment works (POTW)
• Recycle to treatment plant headworks
• Evaporation ponds
• Septic system
• Off-site disposal (non-hazardous, hazardous, radioactive or hazardous and radioactive)
• Land application
• Liquid hazardous waste disposal
• Deep well injection
• Off-gas treatment.

Section C.2, below, describes general design methods and assumptions common across residuals
management options. With two exceptions, subsequent sections describe each of the above
options. Deep well injection is included as an option only in the reverse osmosis/nanofiltration
model and, therefore, is discussed in detail in the chapter relating to that model. Off-gas
treatment is relevant only to aeration technologies and, therefore, is discussed in detail in
chapters relating to aeration models (e.g., packed tower aeration, multi-stage bubble aeration).

83
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit C-1. Technologies and Residuals Generated

Type of Generation
Residuals Generated Technologies
residual Frequency
Intermittent or Adsorptive Media, Anion Exchange, Cation
Spent regenerant or brine1 Liquid
continuous Exchange
Adsorptive Media, Anion Exchange, Biologically
Active Filtration, Biological Treatment, Cation
Spent backwash Liquid Intermittent
Exchange, Greensand Filtration, Granular
Activated Carbon
Adsorptive Media, Anion Exchange, Biologically
Active Filtration, Biological Treatment, Cation
Spent media or resin Solid Intermittent
Exchange, Greensand Filtration, Granular
Activated Carbon
Membrane concentrate Liquid Continuous Reverse Osmosis/Nanofiltration
Spent backwash/tank drain and Low-pressure Membranes
Liquid Intermittent
crossflow (Microfiltration/Ultrafiltration)
Reverse Osmosis/Nanofiltration, Low-pressure
Cleaning waste Liquid Intermittent
Membranes (Microfiltration/Ultrafiltration)
Reverse Osmosis/Nanofiltration, Low-pressure
Spent membrane modules/elements Solid Intermittent
Membranes (Microfiltration/Ultrafiltration)
Used cartridge filters Solid Intermittent Reverse Osmosis/Nanofiltration
Packed Tower Aeration, Multi-stage Bubble
Off-gas Gas Continuous
Aeration, Diffuse Aeration, Tray Aeration
Spent lamps, ballasts and intensity Ultraviolet Disinfection, Ultraviolet Advanced
Solid Intermittent
sensors Oxidation
The chlorine gas, hypochlorite, chlorine dioxide, chloramine, nontreatment, ozone, permanganate addition, phosphate feed,
caustic feed and acid feed models are not shown because no process residuals are generated.
Note:
1. Generated when the technology is used with media regeneration, rather than on a throw away basis.

C.2 General Assumptions

Some of the general assumptions used in developing the costs for management of residuals are
listed below:

• For intermittently generated liquid residuals (e.g., filter backwash), the models calculate
residuals quantities based on the volume of a single generation event (e.g., backwashing
one vessel) and assuming a staggered schedule between generation events (e.g., if vessels
must be backwashed every 48 hours and there are two vessels in operation, the facility
will backwash vessel one at 0 and 48 hours and backwash vessel two at 24 and 72 hours).
• For intermittently generated liquid residuals, flow rates depend on whether flow
equalization is used (e.g., through the use of holding tanks, as described in Section C.3).
• Without flow equalization, the maximum residuals flow rate for intermittently generated
liquid residuals is single generation event volume/event duration.
• With flow equalization, the models assume residuals are released continuously during the
time between generation events. Therefore, the maximum residuals flow rate for
intermittently generated liquid residuals is (single generation event volume/time between
events) x capacity factor. The variable, capacity factor, is present to account for less than
perfect staggering between generation events. The models assume capacity factor equals

84
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

2, but the user can change this assumption on the critical design assumptions sheet of
each model.
• The models size residuals piping, valves and other downstream equipment based on the
maximum flow rates calculated as described above for intermittently generated liquid
residuals and on the maximum continuous flow rate determined by the engineering
models for continuously generated liquid residuals.
• The models assume the length of interconnecting piping between treatment process
equipment and residuals management equipment is equal to 1 times the overall system
building layout length. Like the pipe length assumptions documented in Exhibit 2-10,
this assumption is designed to account for the cost of fittings.
• With a few exceptions (noted in the individual model chapters), the models assume an
additional 40 feet of piping is required for liquid residuals to reach their ultimate
destination (e.g., the discharge point, head of the treatment plant or evaporation pond).
Except when this piping is used to recycle the residual, the models assume this piping is
buried and, therefore, include the cost of excavation, bedding, thrust blocks, backfill and
compaction for the additional pipe length. The user can change the assumption about the
length of the additional residuals piping on critical design assumptions sheet of each
model.
• The models generally assume that total suspended solids (TSS) in the influent water are
completely removed during treatment and accumulate in the residuals generated. This
assumption provides a conservative (high) estimate of the TSS concentration in the
residuals. Assumptions about the concentration of TSS in the influent water vary on a
technology-by-technology basis, but the user can change the assumption on the critical
design assumptions sheet of each model.

C.3 Holding Tanks

The purpose of a holding tank is to equalize the rate of flow at which residuals are released or
discharged. A holding tank may be desirable for intermittently generated liquid residuals that
ultimately are recycled to the treatment plant headworks or discharged to a POTW. The
instantaneous flow of intermittently generated liquid residuals (e.g., filter backwash) during a
generation event can be quite high. The use of a holding tank allows the discharge of these
residuals over the time between generation events, so that the ultimate flow is lower, but more
continuous. When residuals such as filter backwash are recycled to the head of a treatment plant,
recommended engineering practice is that the recycle stream should be no more than 5 percent to
10 percent of total system flow (U.S. EPA, 2002; U.S. EPA, 1996). Flow equalization through
the use of a holding tank may be necessary to meet this generally recommended engineering
practice. It also may be reasonable to include a holding tank for other discharge options (e.g., to
prevent instantaneous flow from overwhelming the capacity of a POTW).

When holding tanks are used for intermittently generated liquid residuals, the models determine
the capacity required as follows: single generation event volume x capacity factor. This capacity
factor is the same variable discussed in Section C.2 and is intended to account for less than
perfect staggering between generation events.

Holding tanks can also be desirable for certain continuously generated liquid residuals (e.g.,
membrane reject) to accommodate variations in flow that occur as influent flow varies. In this

85
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

case, the models determine the capacity required based on a desired detention time. The user can
change this detention time on the critical design assumptions sheet of the appropriate models.

When holding tanks are included, residuals pumps are required to move residuals from the
holding tank to their ultimate destination. The models size these pumps based on maximum
residuals flow rate, as discussed in Section C.2. The models also include maintenance labor,
materials and energy for these pumps in the O&M calculations using the same approach
described for booster pumps in Appendix E.

When holding tanks are used, they can result in the generation of secondary residuals in the form
of solids that settle in the holding tank. The models also allow for the addition of coagulant to the
holding tank to increase the percentage of TSS removed. Users can model this option by
changing the appropriate triggering variable on the critical design assumptions sheet of each
model. When the coagulant addition option is chosen, users also can choose the coagulant used.
Options available (specified on the critical design assumptions sheet) are polymers, ferric
chloride or both ferric chloride and polymers.

By default, holding tanks can be constructed of plastic, fiberglass or steel or they can be open
concrete basins. When the coagulant addition option is chosen, however, the models
automatically assume the tanks will be open concrete basins, to allow for easier solids cleanout.
The models also size the tanks so that a minimum settling time is achieved. When coagulant
addition is chosen, the models also add other required equipment, specifically mixers and dry
feeders or metering pumps.

The following are the model assumptions relevant to solids generation and coagulant addition:

• Without coagulant addition, most models assume that 25 percent of the TSS present in
the residuals is removed in a holding tank 11
• With coagulant addition, this assumption increases to 50 percent
• With coagulant addition, the holding tanks must provide a minimum settling time of 90
minutes
• Coagulant dose is 10 milligrams per liter
• Coagulant sludge production factor is 1 pounds of sludge per pound of polymers added
and 0.99 pounds of sludge per pound of ferric chloride added
• Holding tank solid density is 25 pounds per cubic foot
• Holding tank solids are removed when the solids accumulation reaches 10 percent of tank
capacity.

The user can change each of these assumptions on the critical design assumptions sheet of the
individual models.

11
Exceptions are models, such as anion exchange, that assume low influent solids or include pretreatment filtration
to remove influent solids. These models assume no settling in the holding tank without coagulant addition, because
of the low solids content present (or remaining) in the water being treated.

86
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

C.4 Direct Discharge to Surface Water

Some liquid residuals may be amenable to direct discharge to surface water. Such discharges
require a National Pollutant Discharge Elimination System permit, the costs of which are
included in the add-on costs line item for permits. The only items of capital equipment required
for direct discharges are piping and valves, although the models will include pumps if holding
tanks are used in conjunction with direct discharge (see Section C.3, above).

C.5 Discharge to POTW

Discharge to a POTW is another possible management option for liquid residuals. The discharge
should meet certain pretreatment requirements and must not overwhelm the capacity of the
POTW. The only items of capital equipment required for POTW discharges are piping and
valves, although the models will include pumps if holding tanks are used in conjunction with
POTW discharge (see Section C.3, above).

Discharge to a POTW, however, entails certain charges that are included in the O&M costs of
each model when this discharge option is included. POTW rate structures vary nationwide, but
the most common types of charges are the following:

• Flat fees (e.g., dollars per month).

• Volume-based fees (e.g., dollars per 1,000 gallons discharged).
• TSS-based fees (e.g., dollars per pound of TSS in the discharge if over a certain TSS
concentration). For this fee type, the models assume that the POTW TSS discharge limit
over which a fee is imposed to be 250 parts per million (which is the most common limit
for cities with a limit on TSS).

Individual POTW rate structures can reflect a combination of one or more of these fee types. To
model POTW charges in a way that is nationally representative, the models include all three fee
types and calculate them based on unit charges that represent the average for each fee type based
on data from AWWA (2013). The user can change these average unit charges in the data
extracted from the central WBS cost database. Alternatively, the user can model a specific type
of POTW rate structure by selecting the appropriate option on the critical design assumptions
sheet of each model. The user can indicate which fee types to include (e.g., flat fee only). The
model will then use “typical” unit charges for the selected fee type(s). These “typical” unit
charges, which can be changed in the data extracted from the WBS cost database, reflect the
average including only cities that use that specific fee type (i.e., the average not counting zeros).

C.6 Recycle to Treatment Plant Headworks

Certain liquid residuals can be recycled to the treatment plant headworks provided the system
complies with the backwash recycling rule and the practice does not negatively impact finished
water quality. The recommended engineering practice is that the recycle stream should be no
more than 5 to 10 percent of total system flow (U.S. EPA, 2002; U.S. EPA, 1996). The only
items of capital equipment required for recycling are piping and valves, although a holding tank
(and, therefore, pumps) also would be necessary in most cases to meet the recommendation.

87
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

C.7 Evaporation Ponds

When large quantities of liquid residuals are generated (e.g., spent brine from ion exchange), an
evaporation pond can be an appropriate management method, particularly for facilities in dry
climates. Holding tanks are never necessary with an evaporation pond, even for designs with
intermittent generation frequency, because the design of the pond would provide sufficient
capacity to handle instantaneous flow. A minimum of two cells is recommended to ensure
availability of storage space during cleaning, maintenance or emergency conditions (U.S. EPA,
1987).

When evaporation ponds are selected, the models include the following evaporation pond capital
expenses: excavation, backfill, lining and dike construction. Also, when evaporation ponds are
selected, the models always include the cost of a geotechnical investigation (see Appendix D).
These items are in addition to the pipes and valves required to deliver residuals to the pond. The
models make the following assumptions to design evaporation ponds:

• Arid climate with annual average precipitation of 70 centimeters per year (cm/yr)
• Average annual pan evaporation rate is 180 cm/yr
• Evaporation ratio (which takes into account conversion of pan to lake evaporation rate
and the effect of salinity) of 0.7
• 180 days of storage with no net evaporation
• Evaporation pond safety factor (which accounts for years with below average
evaporation) of 1.1
• Maximum evaporation pond cell area of 5 acres.

The user can change each of these assumptions on the critical design assumptions sheet in each
model that includes the evaporation pond option. If evaporation ponds are selected, the user
should also review the other climate-based assumptions included in the model (e.g., the heating
and cooling requirements on the O&M assumptions sheet) to determine that they are sufficiently
consistent with the assumption of an arid climate that is implicit in the selection of evaporation
ponds as a residuals management method.

The use of an evaporation pond results in the generation of a secondary residual stream in the
form of evaporation pond solids. The models calculate the accumulation of evaporation pond
solids by including all suspended and dissolved solids present in the residuals. The models
assume evaporation pond solids removal frequency of once per year. Users can change this latter
assumption on the critical design assumption sheet of the appropriate models.

C.8 Septic System

Based on comments from peer reviewers, discharge to an in-ground septic tank and drain field (a
septic system) might be an option for some liquid residuals with intermittent generation in small
systems using certain technologies (e.g., adsorptive media, anion exchange). Users selecting this
option should evaluate whether the characteristics of the residuals are appropriate for this type of
discharge. Holding tanks are never necessary with septic systems because the design of the septic
tank would provide sufficient capacity to handle instantaneous flow.

When a septic system is selected, the models include the following capital expenses:

88
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• Septic tanks
• Excavation for septic tanks
• Distribution boxes
• Distribution pipe (perforated polyvinyl chloride)
• Drain field trench excavation
• Drain field gravel.

These items are in addition to the pipes and valves required to deliver residuals to the septic tank.
Also, when a septic system is selected, the models always include the cost of a geotechnical
investigation (see Appendix D). The models make the following assumptions to design septic
systems:

• Minimum septic tank discharge time of 2 days

• Minimum septic tank volume of 1,000 gallons
• Maximum septic tank volume of 100,000 gallons
• Septic tank volume safety factor of 150 percent
• Long-term acceptance rate (a value, based on soil type, used by states/localities to
determine the minimum drain field infiltration area) of 0.5 gallons per day per square foot
• Septic drain field trench width of 4 feet
• Septic drain field trench depth of 4 feet
• Septic drain field trench gravel depth below distribution pipe of 1 foot
• A minimum of two septic drain field trenches
• A maximum septic drain field trench length of 100 feet
• 8 feet between drain field trenches
• Septic drain field trench total gravel depth of 28 inches, based on 1 foot below and 1 foot
above the distribution pipe and a 4 inch pipe diameter
• Septic drain field buffer distance of 10 feet
• Septic tank overexcavation depth of 1 foot above and to each side of the tank
• A maximum of 7 distribution pipe connections per distribution box
• Septic system distribution pipe diameter of 4 inches.

These assumptions are based on values typically found in state and local regulations for septic
systems. The user can change each of these assumptions on the critical design assumptions sheet
of each model that includes the septic system option. The use of a septic system results in the
generation of a secondary residual stream in the form of septic tank solids. The models calculate
the accumulation and disposal cost for these solids using the same assumptions used for holding
tank solids (except that addition of coagulant is not included for septic systems).

C.9 Off-Site Disposal

For solid residuals, including secondary residuals like holding tank solids or evaporation pond
solids, most of the models offer two options: disposal in a hazardous or non-hazardous off-site
landfill. The models do not include disposal in an on-site landfill as an option. This option would
be economically viable only for facilities with an existing on-site landfill—a factor that is highly
site-specific. For these facilities, the cost of this option would be less than that for off-site

89
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

disposal, because it would involve much lower transportation costs. Therefore, the off-site
disposal options available in the models provide a conservative cost estimate for these facilities.

For certain solid residuals, many of the models also offer two additional options: off-site disposal
as a radioactive waste or off-site disposal as a hazardous and radioactive waste. The radioactive
waste disposal options assume that the residuals are classified as low-level radioactive wastes
(LLRW), instead of technologically-enhanced, naturally-occurring radioactive materials
(TENORM). In some cases, TENORM is accepted at traditional non-hazardous or hazardous
waste disposal facility. In such cases, disposal costs would be lower than those at specialized
radioactive waste disposal sites. Therefore, the LLRW disposal costs assumed in the models
provide a conservative cost estimate for cases where residuals might be classified instead as
TENORM.

The models calculate annual disposal costs for non-hazardous solid residuals as follows:

Annual disposal costs = Disposal costs + Transportation costs

where:
Disposal costs = quantity of solids per disposal event (in tons per event) x disposal
frequency (in events per year) x unit cost for non–hazardous waste disposal (in dollars
per ton)
Transportation costs = quantity of solids per disposal event (in tons per event) x disposal
frequency (in events per year) x distance to disposal site (in miles) x unit cost for non–
hazardous waste transportation (in dollars per ton per mile).

The disposal costs for hazardous, radioactive and hazardous radioactive solid residuals are
calculated in a similar fashion. For transportation costs, however, there is a minimum charge per
shipment applied. If transportation costs calculated based on dollars per ton per mile are less than
this minimum, the models calculate transportation costs based on this minimum.

The following are the model assumptions relevant to off-site landfill disposal:

• 10 miles to the nearest non-hazardous waste disposal site

• 200 miles to the nearest hazardous waste disposal site
• 700 miles to the nearest radioactive or hazardous radioactive waste disposal site
• Maximum waste shipment size of 18 tons.

The user can change each of these assumptions on the critical design assumptions sheet of each
model.

C.10 Land Application

When secondary solids (e.g., holding tank solids, evaporation pond solids) are non-hazardous,
most models provide the option of assuming land application instead of landfill disposal. Users
can select this option on the critical design assumptions sheet. When land application is chosen,
the models assume that transportation and disposal costs for the secondary solids are zero,

90
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

although they still include the operator labor costs associated with managing the secondary
solids.

C.11 Liquid Hazardous Waste Disposal

In some site-specific cases, the only viable option for certain liquid residuals (e.g., anion
exchange brine) might be off-site disposal as a hazardous waste. When this option is chosen, the
models automatically include a holding tank, which is required to store the residuals for
shipment. Any solids that settle in the holding tank also are assumed to require hazardous waste
disposal.

The models calculate costs for the liquid hazardous waste disposal option similarly to the off-site
hazardous waste landfill option (e.g., disposal cost + transportation cost, with a minimum charge
per shipment), except that unit costs are different. These unit costs are specific to off-site liquid
hazardous waste disposal, instead of off-site hazardous waste solids landfilling, and expressed in
dollars per gallon or dollars per gallon per mile. The models assume the maximum liquid
hazardous waste shipment size is 6,000 gallons.

C.12 List of Abbreviations and Symbols in this Appendix

cm/yr centimeters per year
LLRW low-level radioactive waste
O&M operating and maintenance
POTW publicly owned treatment works
TENORM technologically-enhanced, naturally-occurring radioactive materials
TSS total suspended solids
WBS work breakdown structure

C.13 References
American Water Works Association (AWWA). 2013. 2012 Water and Wastewater Rate Survey.
Denver, Colorado: AWWA. February.

U.S. Environmental Protection Agency (U.S. EPA). 1987. Dewatering Municipal Wastewater
Sludge. EPA Design Manual. EPA/625/1-87/014. September.

U.S. EPA. 1996. Technology Transfer Handbook: Management of Water Treatment Plant
Residuals. United States Environmental Protection Agency, Office of Research and
Development. EPA 625-R-95-008. April.

U.S. EPA. 2002. Filter Backwash Recycling Rule: Technical Guidance Manual. United States
Environmental Protection Agency, Office of Groundwater and Drinking Water. EPA 816-R-02-
014. December.

91
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Appendix D. Indirect Capital Costs

D.1 Introduction
Indirect capital costs are costs that are not directly related to the treatment technology used or the
amount or quality of the treated water produced, but are associated with the construction and
installation of a treatment technology and appurtenant water intake structures. These costs
represent some of the expenditures required in order to get a technology or the treated water
production process up and running. They include indirect material costs (such as yard piping and
wiring), indirect labor costs (such as process engineering) and indirect burden expenses (such as
administrative costs).

Indirect capital costs included in the work breakdown structure (WBS) models include the
following:

• Mobilization and demobilization

• Architectural fees for treatment building
• Equipment delivery, equipment installation and contractor overhead and profit
• Site work
• Yard piping
• Geotechnical
• Standby power
• Electrical infrastructure
• Instrumentation and control
• Process engineering
• Contingency
• Miscellaneous allowance
• Legal, fiscal and administrative
• Sales tax
• Financing during construction
• Construction management and general contractor overhead
• City index.

The following sections describe each of these indirect cost elements in more detail, address their
effect on capital costs and explain the reasoning behind including them as an additional indirect
capital cost allowance or contingency.

D.2 Mobilization and Demobilization

Mobilization and demobilization costs are costs incurred by the contractor to assemble crews and
equipment onsite and to dismantle semi-permanent and temporary construction facilities once the
job is completed. The types of equipment that may be needed include: backhoes, bulldozers,
front-end loaders, self-propelled scrapers, pavers, pavement rollers, sheeps-foot rollers, rubber
tire rollers, cranes, temporary generators, trucks (e.g., water and fuel trucks) and trailers. In some
construction contracts, mobilization costs also include performance bonds and insurance.

92
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

To estimate mobilization and demobilization costs in the absence of site-specific data, the WBS
models use a multiplication factor of 2 to 5 percent. The models apply this multiplication factor
to direct process costs, building costs and the physical portions of indirect capital costs (site
work, yard piping, geotechnical, standby power, electrical, instrumentation and control and
miscellaneous). Examples of mobilization and demobilization percentages include:

• Buckeye, Arizona Water System Infrastructure Improvements (multiple projects)

Mobilization/Demobilization/Bonds/Insurance = 5 percent (Scoutten, Inc., 2009)
• City of New Port Richey Maytum Water Treatment Plant Modifications,
Mobilization/Demobilization (limit included in bid instructions) = 4 percent (Tampa Bay
Water, 2006)
• Alton Water Works Mobilization = 1 percent (AWWC, 1999)
• Fairfax Water Authority New Intake, Mobilization = 4.6 percent, Demobilization = 1.8
percent, Total = 6.4 percent (FWA, 2003)
• Fairfax Water Authority Trunk Sewer Project Mobilization = 5 percent (FWA, 2003)
• Forest Park Water Treatment Plant, Chalfont, PA = 0.26 percent (Allis, 2005).

The last example, for the Forest Park treatment plant, applied to a retrofit of an existing
conventional filtration facility with a membrane system. The project involved modifications to
existing buildings and treatment basins and the construction of one new building. Since the
project involved less new construction than a greenfield project, the mobilization cost may be
lower than it otherwise would be.

Mobilization/demobilization costs tend to be proportionately higher for smaller projects because

of fixed costs that are the same regardless of project size. For example, if construction requires
use of a large crane, then the mobilization/demobilization cost will be the same regardless of
whether it is onsite for a long time to complete a large construction project or a short time to
complete a small project. Therefore, small projects will most likely have
mobilization/demobilization percentages in the higher end of the range and larger projects will
tend to have values in the lower end of the range. The default values in the WBS models reflect
this type of variation. For small systems with a design flow less than 1 million gallons per day
(MGD), the default mobilization/demobilization factor is 5 percent. For medium systems (design
flows between 1 MGD and 10 MGD), the default factor is 4 percent and for large systems
(design flows above 10 MGD), the default factor is 2 percent. The models make an exception in
the case of small systems that use pre-engineered package treatment plants. Because these
package plants typically are skid-mounted, they require only a short time to install onsite and
should use a minimum of heavy equipment in the process. Therefore, the models assume a
mobilization/demobilization factor of 0 percent for small, pre-engineered package systems. The
user can change this assumption on the indirect assumptions sheet of each model.

Because the installation costs in the models include rental of equipment for installation (see
Section D.4.2), there may be some redundancy between the default mobilization and
demobilization costs and the installation costs (which are included in the model unit costs). The
extent of this redundancy is difficult to determine, but is a potential source of conservatism in
model cost estimates (i.e., the potential redundancy would tend to make model cost results
higher).

93
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

D.3 Architectural Fees for Treatment Building

The architectural fees for the treatment building include the costs of designing the structure and
preparing technical drawings. By convention, the architectural fee also includes the fees for
structural, electrical and mechanical engineering associated with the treatment building
(RSMeans, 2013). Furthermore, the architectural fees include the costs of preparing final
drawings and the tender document package. The building costs in the WBS cost database (see
Appendix B) do not include architectural fees, so the fees are added as an indirect cost. The
models apply the architectural percentage only to treatment building costs, not to other process
costs.

The WBS models use architectural fees from RSMeans (2013), based upon the direct cost of the
building, as shown in Exhibit D-1. The models make an exception in the case of small systems
with a design flow of less than 1 MGD. Because they are typically housed in small, prefabricated
buildings that require a minimum of design and engineering, the models assume no architectural
fee for these small systems. The user can change this assumption on the indirect assumptions
sheet of each model.

Exhibit D-1. Architectural Fees

Building Direct Cost Range Architectural Feea

<$250,000 9.0%
$250,000 to $500,000 8.0%
$500,000 to $1,000,000 7.0%
$1,000,000 to $5,000,000 6.2%
$5,000,000 to $10,000,000 5.3%
$10,000,000 to $50,000,000 4.9%
>$50,000,000 4.5%
a. The architectural fee is a percentage of the direct cost for buildings. It includes a structural engineering fee, as well as
mechanical and electrical engineering fees that are associated with the building.
Source: RSMeans (2013), reference table R011110-10.

D.4 Equipment Delivery, Equipment Installation and Contractor

Overhead and Profit
The equipment unit cost estimates in the WBS database include the cost of equipment delivery,
equipment installation and contractor overhead and profit (O&P). Because these costs are
included in the direct or process costs, the default value of this multiplier in the WBS models is 0
percent. If the user has site-specific or technology-specific data that show delivery, installation or
O&P costs outside of typical ranges, the user can change this factor on the indirect assumptions
sheet of each model to better account for actual installation costs.

The sources of unit cost quotes include manufacturers, vendors, published construction cost data
reference books and peer-reviewed literature. Price quotes for an item vary across sources
because of inherent price variability or product quality differences that are not relevant to overall
performance. The U.S. Environmental Protection Agency (EPA) addressed this source of price
variability by including quotes from multiple vendors in the WBS cost database; the unit costs

94
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

used in the WBS models are simple averages across vendor quotes. Differences also arise
because vendors include different information in price quotes. For example, prices obtained from
RSMeans (2013) include all components needed for installed process costs (i.e., delivered
equipment, installation and O&P costs). Quotes from other sources may not include installation
costs, contractor O&P or transportation costs. Thus, before EPA calculated average costs, all
prices needed to be adjusted to the same installed cost basis. EPA converted costs to this basis by
adding transportation, installation and O&P costs where they were missing from the original unit
price estimates.

D.4.1 Equipment Delivery

Incorporating delivery costs in unit costs that will be used for a national cost analysis is
challenging because of variability in the methods used to assess transportation costs. For
example, transportation costs can be based on a cost per mile, a cost per unit of weight, a cost per
unit of volume, a cost per region or within a radius or a proportion of sales price. EPA developed
standardized transportation cost multipliers that vary by equipment type and size. The type of
multiplier selected for each equipment category is based on a likely method of transportation.

For tanks, vessels and towers, EPA applied a transportation cost based on equipment volume
units (e.g., gallons). For iron and steel tanks, the cost is based on a vendor quote of shipping
costs of $1,000 per 10,000 gallons of tank volume. For fiberglass tanks, the cost is based on a
vendor quote of shipping costs of $600 per 10,000 gallons of tank volume. EPA included
minimum and maximum transportation costs for tanks, vessels and towers. For steel equipment,
the minimum transportation cost is $500, which applies to all items with volumes below 5,000
gallons, and the maximum transportation cost is $5,000, which applies to all items with volumes
over 50,000 gallons. For plastic/fiberglass equipment, the minimum transportation cost is $500,
which applies to all items with volumes less than 10,000 gallons, and the maximum
transportation cost is $3,000, which applies to all items with volumes over 50,000 gallons. For
very small plastic/fiberglass equipment, EPA used an alternative minimum shipping charge of
$100, which applies to all items with volumes less than 100 gallons.

To estimate transportation costs for pipe, EPA calculated delivery costs per linear foot of pipe
using vendor delivery cost estimates and linear feet/truck load estimates. EPA averaged two
vendor delivery estimates for 30-inch and 48-inch American Water Works Association C200
steel pipe to obtain an estimate of $197.75 for a truckload of pipe. Information obtained from
vendors was used to estimate the number of linear feet of each size pipe that could fit in a
truckload.

For valves, pumps, blowers and mixers, EPA developed transportation cost estimates based on
equipment weight and costs for “less than a load” (LTL) shipments obtained from vendors. The
estimates assume an average delivery distance of 100 miles. For shipping cost estimation
purposes, average weights were assumed for the small, medium and large sizes of valves, pumps,
blowers and mixers. The assigned weights (which are based on the actual weights of valves,
pumps, blowers and mixers for which EPA received vendor quotes) are as follows:

• Small steel valves ~ 30 pounds

• Medium steel valves ~ 80 pounds
• Large steel valves ~ 400 pounds

95
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

• Small pumps / blowers ~ 100 pounds

• Medium pumps / blowers ~ 300 pounds
• Large pumps / blowers ~ 600 pounds
• Small mixers ~ 50 pounds
• Medium mixers ~ 100 pounds
• Large mixers ~ 400 pounds.

Since the density of 304 stainless steel is approximately 5.6 times greater than the density of
polyvinyl chloride, the following weights were assigned to plastic valves:

• Small plastic valve ~ 5 pounds

• Medium plastic valves ~ 15 pounds
• Large plastic valves ~ 70 pounds.

EPA rounded shipping costs to the closest $10 increment. Exhibit D-2 provides the weight
categories and LTL costs for valves, pumps, blowers and mixers, along with a complete
summary of transportation cost methods for all categories of equipment.

EPA assumed a 5 percent markup on miscellaneous equipment and filter components for
membrane systems. For system control components, EPA assumed no transportation costs,
because the vendors contacted did not charge for shipping on large orders (i.e., greater than
$300). Transportation costs for chemicals, resins and filter media are averages of delivery costs
obtained from vendors. EPA used shipping rates for standard service from a vendor for
instrumentation transportation costs; the vendor uses fixed shipping rates that vary according to
the equipment price.

D.4.2 Installation, Overhead and Profit

EPA incorporated installation and O&P costs using multipliers derived from RSMeans (2013)
cost data. RSMeans provides complete installed cost estimates for the unit costs in its database.
The following cost components are reported for each unit cost:

• Bare material costs, including delivery

• Installation labor, materials and any rental cost for installation equipment
• Overhead for installing contractor (i.e., labor and business overhead costs)
• Profit for installing contractor (i.e., a 10 percent rate of profit charged on materials,
installation and overhead costs).

96
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit D-2. Transportation Cost Estimation Methods

Equipment Category Transportation Costs
$1,000 per 10,000 gallons of volume for each tank. There is a minimum shipping
Vessels, Tanks, Towers – steel
charge of $500 and a maximum of $5,000.
$600 per 10,000 gallons of volume for each tank. There is a minimum shipping charge
Vessels, Tanks, Towers –
of $500 and a maximum of $3,000. Very small tanks (less than 100 gallons) have an
plastic/fiberglass
alternate minimum shipping charge of $100.
Varies by pipe diameter and material of construction.
Pipes Plastic pipes: range is $0.07–$41.20 per 100 linear feet.
Iron and steel pipes: range is $6.46–$494 per 100 linear feet.
Valves-steel/iron Small valves: $30.00 (1"–4" diameter)
Weight Class 60 Medium valves: $80.00 (5"–9" diameter)
LTL rate = $101.45/100 lb Large valves: $400.00 (10"+ diameter)
Valves-plastic Small valves: $10.00 (1"–3" diameter)
Weight Class 70 – plastics Medium valves: $20.00 (4"–6" diameter)
LTL rate = $115.41/ 100 lb Large valves: $80.00 (>6" diameter)
Pumps and blowers Small units: $130.00 (0–50 gpm)
Weight Class 85 Medium units: $400.00 (51–300 gpm)
LTL rate = $132.08/100 lb Large units: $790.00 (>300 gpm)
Mixers Small mixers: $70.00 (includes mounted and portable mixers)
Weight Class 85 Medium mixers: $130.00 (includes inline and static mixers)
LTL rate = $132.08/100 lb Large mixers: $400.00 (includes turbine, rapid, flocculant, impeller mixers)
Miscellaneous Equipment 5% of equipment cost
System Controls None
$0.22/lb for hazardous materials
$0. 27/lb for filter media and resins
Chemicals, Resins and Filter Media $0.18/lb for 150 lb chlorine cylinders
$0.24/lb for 1 ton chlorine cylinders
$0.06/lb for all other chemicals
RO/NF and MF/UF Skids and
5% of equipment cost
Equipment
Instrumentation Varies with cost of equipment. Range is $9.95 to $104.35 per unit of equipment.
lb = pound; gpm = gallons per minute; “ = inch; RO/NF = reverse osmosis/nanofiltration; MF/UF = microfiltration/ultrafiltration

These component cost data provide enough information to calculate adjustment factors that can
be applied to price quotes that exclude installation and O&P costs. By dividing total unit cost,
which includes all components, by bare material cost including delivery, EPA obtained
adjustment factors for several types of equipment in the WBS cost database. For example, if the
bare material cost, including delivery, for an item of equipment is $1.00 and the total unit cost is
$1.78, then the adjustment factor is 1.78. When unit costs obtained for the database did not
include installation, overhead and profit (as is typical when obtaining costs from manufacturers),
EPA applied these adjustment factors to escalate the unit costs so that they represented the full
installed cost. For example, if a manufacturer’s price for a 20,000 gallon steel tank was $25,000,
EPA would first add delivery cost ($1,000 per 10,000 gallons capacity, as described in Section
D.4.1), resulting in a cost with delivery of $27,000. EPA would then multiply that cost by the

97
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

appropriate adjustment factor (for instance, 1.17) to obtain a complete unit cost—that is, the total
unit cost in this example would be ($25,000 + $2,000) × 1.17 = $31,500.

Most of the installation and O&P multipliers in the WBS cost database fall between 1.03 and
1.73, with an average around 1.36.

D.5 Site Work

Every construction site requires a certain amount of site preparation and finish work. Site work
costs include site preparation, excavation and backfilling, temporary and permanent road
construction, retaining wall construction, final grading, landscaping, parking lots, fencing, storm
water control structures, yard structures, site cleanup, waste disposal and utilities.

Estimating the site work cost based on a factor applied to the direct capital cost is an approach
commonly used when detailed information about the site plan is not known. Under this approach,
site work costs are typically estimated between 5 and 15 percent of the direct capital costs,
depending on project size and scope.

Site work costs vary directly with the land area requirement. The WBS models generate land
area estimates, which allows the models to use an alternative cost estimation approach based on
total project land area instead of total project costs. RSMeans (2013) provides an analysis of
actual reported project and component costs for different types of construction. Of the many
building categories reported in the summary database, the “factory” category best fits the scope
of construction associated with drinking water treatment plants. Therefore, the models use the
national average median project cost for site work at factories from RSMeans (2013). The WBS
cost database automatically updates this unit cost to current year dollars using the Engineering
News-Record (ENR) Building Cost Index (see Chapter 2). The models compute a site work cost
based on this unit cost and the total project land area, excluding land used for residuals holding
lagoons and evaporation ponds. Since the models include the cost of excavation and backfill for
these facilities, there is no need to include them in the site work calculations.

EPA believes that using an approach based on land area instead of direct process costs provides a
better estimate of site work costs because the unit costs from RSMeans (2013) are primarily
based on quantities of area and earthwork volume. Furthermore, this approach is less sensitive to
cost fluctuations caused by high cost equipment—the site work cost for a 0.5-acre project site
will be the same regardless of whether the treatment building houses chemical addition or a
membrane filtration process. This is particularly important because expensive, advanced
treatment technologies often have smaller footprints than lower-cost, conventional technologies
such as conventional filtration. Basing site work costs on process costs will tend to overstate site
work costs for such advanced technologies.

Although the default site work cost in the WBS models reflects a median value, the user can
enter a different rate in the data extracted from the WBS cost database based on site-specific
conditions. A higher cost factor should be entered for projects where the site conditions may
require higher-than-average site work costs (e.g., a site with steep terrain that may require
retaining walls). Conversely, a lower rate should be entered for projects where the site conditions
may require lower-than-average site work costs (e.g., a site where little grading is needed and

98
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

where requirements for infrastructure and other site improvement are minimal or where portions
are already in-place).

D.6 Yard Piping

Yard piping costs reflect the costs to install piping for untreated, partially treated and treated
water to and from the site, between new treatment plant buildings or between existing and new
treatment units. It does not include piping of treatment residuals to a residuals treatment system,
to disposal in a sewer or to a direct discharge connection; those costs are included as explicit
capital cost line items in the relevant WBS technology models, as discussed in Appendix C.

Yard piping costs include the following components:

• Trench excavation, backfill and pipe bedding

• Piping from the boundary of the building buffer zone to and from the building inlet and
building outlet and in between buildings that house water treatment components
• Optionally, piping from the water source to the property boundary and piping from the
property boundary to the distribution system connection
• Thrust blocks.

The sections below describe each of these components.

D.6.1 Trench Excavation, Backfill and Pipe Bedding

Costs of pipe contained in the WBS cost database are installed costs for aboveground pipes
within the treatment facility. Yard piping generally is installed below ground. Therefore, yard
piping entails additional costs. These costs include trench excavation costs, bedding costs,
backfill costs and thrust block costs (discussed in Section D.6.3).

Technology land area requirements are calculated on a basis of starting with a square building
with the required footprint and adding a non-fire buffer (10 feet) on three sides of each building
and a fire buffer (40 feet) on the fourth side. The general configuration assumption is that the fire
buffer will be located along the front side and the distance between buildings will be two times
the non-fire buffer distance (20 feet) and, therefore, yard piping will not cross the fire buffer
area. Thus, the minimum initial trench length is 20 feet (10 feet at inlet and 10 feet at outlet) for a
system with one building or 30 feet (20 feet inlet and outlet and 10 feet between buildings) for a
system with two buildings. Since the inlet and outlet piping may not always line up and may
extend inside the building perimeter, an offset distance is added to the 10 foot buffer distance
based on the building size. The offset distance is assumed to be ¼ the length of one side of the
building footprint (based on square root of total building footprint).

The models assume yard piping will be buried with the top of the pipe set at or below the local
frost depth. Where frost depth is less than 30 inches, a minimum depth of 30 inches is assumed
to provide a protective cover. The default frost depth is 38 inches and corresponds to the frost
depth in St. Louis. Users can change the frost depth on the indirect assumptions sheet of each
model based, for example, on the climate data for a selected city in the climate database
(AFCCC, 2000). Trench depth also incorporates the pipe diameter and the bedding depth, which
the models assume to be 6 inches below the bottom of the pipe. This default value is sufficient to

99
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

approximate bedding requirements for large size pipes laid in soils where bedding is necessary.
The user has the option of changing the default value on the indirect assumptions sheet of each
model.

Trench width is equal to the pipe diameter plus 1 foot on either side. Trench volume is based on
the calculated trench length times the trench cross-sectional area, which incorporates trench
width and depth and assumes sloped trench sides, with an angle of 45 degrees (expressed in
radians on the indirect assumptions sheet). Excavation and backfill costs are based on total
trench volume plus thrust block volume and the unit cost for excavation and backfill. Although
backfill quantities are generally smaller than excavation quantities, they are assumed to be the
same in the WBS models. This approach is assumed to cover to the cost of backfill and the cost
of spreading or hauling excess soil off site. Pipe bedding volume accounts for the bedding depth,
incorporating additional volume to account for the sloped sides of the trench and the assumption
that the bedding covers 25 percent of the pipe diameter. The user can change this latter
assumption on the indirect assumptions sheet of each model.

D.6.2 Piping
The basic assumptions for yard piping from the boundary of building buffer zone to and from the
building inlet and building outlet and in between buildings are:

• Pipe length will be equal to trench length plus two times the trench depth.
• Pipe costs will be based on an equivalent pipe length, which will include an additional
length to account for cost of fittings (e.g., elbows). The equivalent length will be equal to
two times the pipe length, using the same factor used for process piping within the
buildings (see Section 2.4).
• Yard piping costs do not include valves.
• Piping materials, diameter and unit cost are the same as those selected in the treatment
model for inlet and outlet piping within the building.

In addition, the indirect assumptions sheet in each model contains an optional assumption for the
length of yard piping from the water source and another optional assumption for the desired
length of yard piping to the distribution system. Therefore, if the technology is not the initial step
in the treatment train, the default value length of pipe from the water source should be 0 feet,
because there is already a pipe from the water source to the existing facility. Similarly, if the
technology is not the last technology in the treatment train, then the default value should be 0
feet. As a default, these assumptions are set to zero.

D.6.3 Thrust Blocks

Yard piping costs include concrete thrust blocks to hold small pipe elbows and other fittings in
place. The basis of the thrust block volume calculation is thrust force in pounds. The models
derive thrust force using a lookup table based on pipe diameter. Users can modify the thrust
force assumptions by editing the engineering lookup table extracted from the WBS cost database.
The values in the lookup table assume a pipe test pressure of 150 pounds per square inch and a
pipe elbow angle of 90 degrees and account for block weight. Although both vertical and
horizontal elbows are expected in every pipe-laying job, the thrust block calculations assume
horizontal thrust blocks.

100
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Using the data from the thrust force lookup table, the models calculate bearing surface area based
on a conservative approach found in U.S. Army Corps of Engineers guidance (U.S. ACOE,
1992). The calculation is:

Area = 1.5*T/(Soil Density*Kp*Depth*R)

where:
1.5 is a safety factor, which is typical for thrust block design
T is the thrust force required, derived as discussed above
Soil Density is the minimum soil bulk density, which the models assume is 1.55 grams
per cubic centimeter (97 pounds per cubic foot) consistent with loamy sand, which is also
on the lower end of the range for sandy soils (1.5–1.8) and the upper end of the range for
silty clay (1.4–1.5) (MN NRSC, 2003)
Kp is the coefficient of passive pressure, which the models assume to be 3, based on an
internal angle of friction of the soil (phi) of 30 degrees
Depth is the depth to bottom of the block, which the models calculate based on trench
depth and pipe diameter
R is a reduction factor of 0.467, based on phi of 30 degrees and a vertical bearing surface
(CADOT, 2001, Figure 8).

Users can adjust soil density, Kp and R on the indirect assumption sheet of each model. Note that
this approach is conservative in that it considers only the bearing force of the vertical surface,
which is perpendicular to the thrust force, and ignores the frictional force exerted on the bottom
surface of the block. Use of deeper trench depths will result in lower thrust block costs.

D.7 Geotechnical Investigation

Construction cost estimates generally include a geotechnical allowance to provide for
investigation of subsurface conditions. Subsurface conditions can affect the foundation design
and construction technique. For example, a high groundwater table or soft substrate may require
special construction techniques such as piles and dewatering. Thus, the actual costs of addressing
subsurface conditions are site specific and can vary considerably. In addition, where a system is
adding treatment technology to a site with existing structures and, therefore, the site already has
an existing geotechnical investigation, additional geotechnical investigation may not be required.
To account for these variations, the models include assumptions that allow the user to select
whether geotechnical investigation costs should be included for low, medium and high cost
estimates. The default values for these assumptions include geotechnical costs only for high-cost
systems. However, the models always include geotechnical costs (regardless of the value
selected for these assumptions) when certain components are included in the technology design,
such as septic systems, evaporation ponds or below-grade structures like basins.

Geotechnical investigations can be as simple as digging trenches or test pits to determine the soil
conditions underlying small, lightweight structures. For larger, heavier structures, site
investigations generally involve drilling boreholes to extract samples of rock or soil for further
study. Cost estimates in the WBS models reflect either test pit costs or borehole costs, depending
on the building footprint size. For footprints of 2,000 ft2 or smaller, the WBS models have costs

101
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

based on hand digging test pits. All larger structures have costs based on the costs of drilling
boreholes. The following sections describe the method for estimating costs for each approach.

D.7.1 Borehole Cost Analysis

The cost analysis for drilling boreholes includes preparation activities (e.g., staking the field) and
actual drilling. Thus, a key cost driver is the number of boreholes needed. An additional factor is
the required drilling depth.

For a large industrial building, a borehole should be drilled at the expected location for each
column foundation and at locations where concentrated loads are expected to occur such as under
tanks and heavy equipment. The models assume four boreholes is reasonable for structures in the
range of 2,000 to 4,000 ft2. For larger structures, the models assume an additional borehole for
every 1,000 ft2 in additional space. Thus, the requirement for structures in the range of 4,001 to
5,000 ft2 is five boreholes. This approach is based on the assumption that column footings are
spaced approximately 32 feet apart.

Drilling depth depends on a structure’s weight and existing knowledge of subsurface conditions.
Nevertheless, a rough criterion used to develop WBS cost estimates is that boreholes should
penetrate at least 1.5 times the width of the footings below the lowest portion of the footing
(Krynine and Judd, 1957). The lowest portion of the footing must be below the frost line, which
ranges from almost 0 feet to more than 5 feet in the continental United States. The WBS models
assume a frost line depth of 38 inches, an additional safety depth of 22 inches and a footing
width of 3 feet to obtain a minimum borehole depth of approximately 10 feet (5 feet + 1.5 x 3
feet).

EPA selected three different boring depths to represent a range of geologic conditions and
building bearing loads. A boring depth of 10 feet applies to relatively light structures in areas
where the soil conditions are predictable without any expectation of deeper strata that exhibit
poor shear strength. A boring depth of 25 feet applies to moderately heavy structures in areas
where subsurface conditions are less well defined, but no severe conditions are expected and
where underground structures, such as basins, as deep as 20 feet need to be constructed.
Similarly, a boring depth of 50 feet deep applies to heavy structures in areas where extreme or
unknown subsurface conditions (such as strata with poor shear strength) may exist.

EPA developed cost estimates based on cost data for drilling activities that use a truck-mounted,
2.5-inch auger with casing and sampling from RSMeans (2013). Exhibit D-3 identifies the cost
elements included in borehole drilling. The WBS cost database automatically updates the unit
costs for these elements to current year dollars using the ENR Construction Cost Index (see
Chapter 2). Costs are applied based on the selected borehole depth and total structure area
rounded up to the nearest thousand ft2.

102
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit D-3. Cost Elements Included in Borehole Drilling

Item
• Borings, initial field stake out and determination of elevations
• Borings, drawings showing boring details
• Borings, report and recommendations from professional engineer
• Borings, mobilization and demobilization, minimum
• Borings, drill rig and crew with truck mounted auger (output 55 feet/day)
• Borings, cased borings in earth, with samples, 2.5-inch diameter.
Source: RSMeans, 2013, 02 32 13.10-0200.

D.7.2 Test Pit Cost Analysis

For smaller, less expensive buildings, boreholes are less cost effective compared to test pits or
trenches that can be dug by hand or by using earth moving equipment if it is already available at
the site. Because geotechnical investigations may precede site work, excavating equipment may
be available to dig test pits. Therefore, for small buildings, the models use hand-dug test pits as
the basis for costs. The models assume one pit for buildings up to 1,000 ft2 and two pits for
buildings of 1,001 to 2,000 ft2.

Pit widths range from 4 feet by 4 feet to 6 feet by 8 feet (Krynine and Judd, 1957). Because this
test method is limited to small buildings, the models assume pits that are 4 feet by 4 feet wide.
Pit depth of 7 feet is based on a footing width of 2 feet and a frost depth of 5 feet (5 feet + 1.5 x 2
feet). The unit excavation and backfill costs are based on data from RSMeans (2013) for hand
dug pits in heavy soil. The cost of surveying and the soil sample evaluation report and
recommendation from a Professional Engineer are assumed to be the same as for borings.

D.8 Standby Power

A new treatment facility sometimes requires a standby power source that can produce enough
energy to operate the facility in the event of an electricity outage. Thus, the power rating or
capacity of the standby generator should be sufficient to power critical operating components at
the rated maximum flow capacity of the equipment (i.e., the design flow). Critical components in
a treatment plant include pumps, lighting and ventilation. In addition, standby power can be
required to provide space heating (if an electrical resistance heater or heat pump is used) and/or
cooling in the event of a power outage. As a default, the WBS models do not include heating or
cooling in their estimate of standby power requirements. The user can change the assumptions
about inclusion of heating and/or cooling in standby power on the indirect assumptions sheet of
each model. 12

Also as a default, the models do not include standby power at all for small systems with a design
flow of less than 1 MGD. These small systems typically operate for only a few hours each day,
placing water in storage for use during the rest of the day. This operating procedure means small
systems can handle short term power outages simply by postponing their operating hours,

12
Note that if the assumption about including heating in standby power is changed, heating requirements will only
be included in standby power if an electrical resistance heater or heat pump is used, because the other heating
options (e.g., natural gas heat) do not use electricity.

103
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

without the need for standby power systems. The user can change the assumption about
including standby power for small systems on the indirect assumptions sheet of each model.

The generation capacity requirement for critical systems is based on the maximum daily load,
which is the potential energy demand to meet production at the design flow rate. Since the
energy requirements calculated in the models are based on continuous operation (24 hours/day
and 365 days/year), the maximum power requirement in kilowatts (KW) can be estimated using
the following equation:

Power requirement for critical operating components (KW) =

[annual power use by critical operating components (MWh/yr) / 365 (day/yr) / 24 (hr/day)] *
1,000 (KW/MW)

where:
hr = hours
MW = megawatt
MWh/yr = megawatt hours per year
yr = years

Standby power costs primarily comprise equipment purchase (e.g., a generator) and installation.
Additional costs include fuel purchase and storage. Annual fuel costs for standby power
generation are hard to estimate or predict, given the unpredictable nature of using the standby
power generator. Typical standby generators consist of diesel engine powered generators
(NREL, 2003). Installation costs include provisions for a foundation, fuel storage and louvered
housings for larger systems. For the diesel generators typically used for standby power, EPA
used installed unit costs from RSMeans (2009a). The WBS cost database automatically updates
these unit costs to current year dollars using the Producer Price Index from the Bureau of Labor
Statistics for motors, generators and motor generator sets (see Chapter 2). The models multiply
the appropriate unit cost, which users can change by modifying the data extracted from the WBS
cost database, by the calculated standby power requirement in KW, after applying a minimum
requirement of 1.5 KW (based on the smallest available standby power generator).

D.9 Electrical
The electrical cost allowance in a construction cost estimate will primarily account for electric
wiring inside structures, such as wiring for motors, duct banks, motor control centers, relays and
lighting. The unit costs for buildings in the WBS models (see Appendix B) already incorporate
general building electrical, such as building wiring and lighting fixtures and electrical
engineering associated with those components. In addition, certain electrical costs (motor/drive
controllers, variable frequency drives and switches) are included in direct costs for system
controls and pumps. Technologies with significant process equipment located outside include an
electrical enclosure as an explicit line item. Thus, the indirect cost electrical allowance only
accounts for additional electrical equipment associated with the treatment facility, including
outdoor lighting, yard wiring, switchgear, transformers and miscellaneous wiring. Yard wiring
consists primarily of the infrastructure that connects a new treatment facility to the power grid
and, if necessary, converts voltage.

104
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Typical electrical percentages include:

• Building electrical as a percentage of building cost = 7.7 to 13.0 percent, depending on

building size and quality (Association for the Advancement of Cost Engineering
International [AACEI] building cost model results)
• Seymour, Indiana electrical costs as a percentage of non-electrical process costs = 12.1
percent (AWWC, 2001a)
• St. Joseph, Missouri electrical costs as a percentage of non-electrical process costs = 8.7
percent (AWWC, 2001b).

Based on these data, the electrical percentage in the WBS models is 10 percent as a default.
Users can change this assumption on the indirect assumptions sheet of each model.

D.10 Instrumentation and Control

Instrumentation and control (I&C) costs include a facility control system and software to operate
the system. The WBS models include detailed process cost estimates for instrumentation and
control, as described in Appendix A. Therefore, the default value for I&C on the indirect
assumptions sheet is 0 percent. This line item remains among the indirect costs on the output
sheet of each model, however, to allow the user to incorporate any site-specific or technology-
specific data that cannot be accommodated by altering the I&C design assumptions in the WBS
models.

D.11 Process Engineering

Process engineering costs include treatment process engineering, unit operation construction
supervision, travel, system start-up engineering, operating and maintenance manual development
and production of record drawings. Process engineering as a percent of installed process capital
cost ranges from 5 to 20 percent. For example, Brayton Point Power Plant Water Works process
engineering costs were estimated at 8 percent of installed capital costs (Stone and Webster,
2001).

The ratio of process engineering to installed process capital cost varies based on system size and
the complexity of the treatment process. In particular, engineering cost as a percentage of process
cost tends to decrease as the size of the treatment plant increases. The default values in the WBS
models reflect this pattern: 8, 12 and 20 percent for large, medium and small systems,
respectively. The WBS models apply these percentages to installed process costs, but not
building costs, because structural, mechanical and electrical engineering fees are included in the
architectural fee (Section D.3).

The process engineering percentages at 13 EPA demonstration sites for low-flow packaged
systems ranged from 20 to 80 percent, with a mean of 36 percent (U.S. EPA, 2004). These
percentages, however, also include permitting and administrative costs. Because these costs are
separate line items in the WBS models, these percentages overstate stand-alone process
engineering costs. Furthermore, engineering costs can be higher for technologies in the
demonstration phase than for those in wide use. Therefore, EPA retained its assumption of 20
percent process engineering cost for small systems.

105
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

D.12 Contingency Cost

Contingency cost reflects the degree of risk that management assigns to a project. This cost
should reflect the statistical probability of additional project costs because of uncertainties and
unlikely or unforeseen events (AACEI, 1996). These unforeseeable additional costs to the project
occur because of changes in design, materials, construction methods and/or project schedule.
Contingency reflects a judgment by project management or bidders to account for unforeseeable
costs, thereby avoiding cost overruns. Contingency costs are included as part of a construction
contract allowing the contractor to be paid extra upon authorization of design and construction
changes by the project engineer (AACEI, 1996).

The risk of additional unforeseen costs associated with construction projects tends to vary with
project size and complexity. Therefore, EPA developed contingency factors using both project
size (i.e., total direct cost) and complexity (i.e., the technology being modeled) as input
variables. Ideally, a contingency cost estimate is based on statistical data or experience from
similar projects. By their nature, however, contingency costs are site specific and difficult to
predict; two estimators may recommend different contingency budgets for the same project
(Burger, 2003). EPA examined recommended contingency values, tabulated by project size,
from an economic analysis of water services (GeoEconomics Associates Inc., 2002). The
recommendations are presented in Exhibit D-4. These contingency rates, which range from 2 to
10 percent, are applied to the base costs (i.e., direct costs) to derive contingency cost. These rates
apply to projects of low to average complexity. Water treatment construction projects typically
fall into this category, depending on the technology being installed.

Exhibit D-4. Recommended Contingency Rates from an Economic Analysis of

Water Services
Project Base (Direct) Cost Contingency as a Percent of Base Costs
Up to $100,000 10%
$100,000 to $500,000 8%
$500,000 to $1,500,000 6%
$1,500,000 to $3,000,000 4%
Over $3,000,000 2%
Source: GeoEconomics Associates Inc. (2002)

The WBS models would ideally include only the part of a contingency budget that is actually
spent, rather than the total amount budgeted. EPA therefore considered a Construction Industry
Institute (2001) study, which included both budget estimates and actual spending for contingency
in a series of heavy construction projects. Exhibit D-5 presents the relevant results. The factors
are expressed as a percentage of the total budget, rather than direct costs. These projects are not
limited to water treatment systems and include a variety of heavy construction projects. The data
in Exhibit D-5 show that, with the exception of very large projects (those with total project costs
of over $100 million), the contingency cost tends to decrease as project size increased. The
average contingency factor decreases from 6 to 4 percent before increasing to 7 percent for very
large projects. Such very large projects are generally beyond the size of projects that can be
modeled using the WBS models. Exhibit D-5 also shows that unforeseen problems during

106
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

construction tend to account for a higher share of contingency cost than design or procurement
problems.

Exhibit D-5. Average Contingency Costs in Budgets for Heavy Industrial Projects
Contingency Costs
Total Contingency Contingency Final Incurred by Project Phase Contingency
Project Cost Budget (% of Project (Design / Procurement / Incurred /
Size Budget Estimate budget) Cost Construction) Budgeted
<$15 8.09 0.46 6% $7.76 0.34 (0.04 / 0.10 / 0.20) 74%
$15–$50 30.22 1.55 5% $29.51 1.15 (0.20 / 0.30 / 0.65) 74%
$50–$100 70.70 3.09 4% $68.19 2.24 (0.25 / 0.83 / 1.16) 72%
>$100 214.02 15.56 7% $206.50 13.63 (2.00 / 4.24 / 7.39) 87%
All costs are in millions of dollars. Incurred contingency costs exclude excludes three phases: Project Planning Phase,
Demolition and Start Up.
Source: CII (2001)

The contingency factors in Exhibit D-5 are higher than the recommended values in Exhibit D-4.
Because Exhibit D-5 data is empirical and the basis for the estimates in Exhibit D-4 is not clear,
EPA based its contingency factors in the WBS models primarily on the values in Exhibit D-5,
but incorporated additional price categories below $15 million with contingency factors above 6
percent. To create the contingency factors, EPA first converted the figures in Exhibit D-5, which
are expressed as percentages of a total budget, to markups. For instance, if the contingency
budget is 7 percent of a total budget, it represents a markup of 7 / (100 - 7) percent = 7.5 percent.
EPA modified the markups by a factor of 0.77, which is the average ratio of incurred to budgeted
contingency costs in Exhibit D-5. Exhibit D-6 presents the resulting base contingency factors.
These represent the contingency or risk prior to consideration of technology complexity.

Exhibit D-6. WBS Model Contingency Factors Prior to Consideration of

Technology Complexity

Project Direct Cost Range Base Contingency Factor

<$500,000 6.7%
$500,000 to $3,000,000 5.8%
$3,000,000 to $15,000,000 4.9%*
$15,000,000 to $50,000,000 4.1%*
$50,000,000 to $100,000,000 3.2%*
>$100,000,000 5.8%*
* Percentages based on CII-Benchmarking & Metrics Analysis Results (CII, 2001).

While there are techniques and computer programs designed to estimate contingency factors for
large projects based on construction activity risk simulation, the engineering costing literature
and the example projects EPA reviewed do not provide specific quantitative guidance regarding
the effect of project complexity on contingency costs. Nevertheless, the anecdotal evidence
suggests that risks (and, therefore, contingency costs) increase when project complexity
increases.

Among the WBS technologies, project complexity depends on the type of technology employed
and the general degree of experience with the technology as it will be applied. Well-established

107
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

technologies, which have a depth of construction and technology installation and operational
history under a variety of conditions, are expected to have low risk with respect to unforeseen
problems during construction and installation. Recently developed technologies or ones that have
had limited application to a variety of water quality conditions and project sizes (or to the
conditions at the project in question) are expected to have a higher degree of risk.

To account for differences in contingency values associated with technology type and project
complexity, EPA identified four categories of project complexity and assigned multipliers that
the models use to adjust the contingency factors (up or down) from Exhibit D-6:

• Low complexity = base contingency factor x 0.5

• Average complexity = base contingency factor x 1.0
• High complexity = base contingency factor x 1.5
• Very high complexity = base contingency factor x 2.0.

Thus, for each technology, the applied contingency factor combines the effects of project size
and technology complexity to obtain the project specific contingency factor. EPA assigned a
project complexity category to each WBS technology based on general knowledge and the
application history of the technology to drinking water treatment. Exhibit D-7 shows this default
complexity category assignment. The user can change these values on the indirect assumptions
sheet of each model if specific knowledge of the technology and its expected performance under
the site-specific conditions warrant such a change.

The WBS models assume that contingency costs are incurred only in high cost scenarios (see
Section 2.3). For low and medium cost estimates, the models assume construction is completed
with a minimum of unforeseen site-specific costs and, therefore, that none of the contingency
budget is incurred. Users can change this assumption on the indirect assumptions sheet of each
model.

108
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit D-7. WBS Default Complexity Factors by Technology

Risk Level Assigned to Default Complexity
Technology Technology Factor
Acid Feed Low 0.5
Cation Exchange Low 0.5
Caustic Feed Low 0.5
Chloramine Low 0.5
Nontreatment Options Low 0.5
Phosphate Feed Low 0.5
Permanganate Addition Low 0.5
Granular Activated Carbon Average 1
Chlorine Gas Average 1
Diffuse Aeration Average 1
Packed Tower Aeration Average 1
Adsorptive Media High 1.5
Anion Exchange High 1.5
Biological Treatment High 1.5
Biologically Active Filtration High 1.5
Chlorine Dioxide High 1.5
Low-pressure Membranes (Microfiltration/Ultrafiltration) High 1.5
Greensand Filtration High 1.5
Hypochlorite High 1.5
Multi-stage Bubble Aeration High 1.5
Ozone High 1.5
Reverse Osmosis/Nanofiltration High 1.5
Tray Aeration High 1.5
Ultraviolet Advanced Oxidation Processes Very high 2
Ultraviolet Disinfection Very high 2

D.13 Miscellaneous Allowance

In a cost estimate for a construction project, an allowance may be included for conditions or
events that the estimator can anticipate, but whose cost is not known with any degree of
certainty. If, for example, the site is expected to have contaminated soil that may require
remediation, the allowance will incorporate the resulting costs. An allowance differs from a
contingency cost, which provides contract coverage for unpredictable conditions. The allowance
funds account for anticipated additional costs that should become apparent at a later stage of the
project (for example, upon completion of the site investigation activities and the detailed
engineering design). Much of this cost is associated with knowledge of site-related conditions.

In a national average cost estimate such as the one that the WBS models generate, it is not
possible to allow for the specific conditions associated with any given site. However, the models
include an allowance line item to simulate an average effect due to such conditions. The line also

109
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

accounts for the level of detail in the WBS design, since the models do not include all minor
process components.

The WBS models assume a miscellaneous allowance of 10 percent as a default. Since the
allowance addresses a modeling uncertainty, there is little guidance available from the cost
estimation literature. Instead, the assumption must be validated by comparing WBS output to
actual water treatment facility construction costs. Users can change the miscellaneous allowance
on the indirect assumptions sheet of each model.

D.14 Legal, Fiscal and Administrative

This category includes project management, accounting and administrative activities related to
the project, excluding permitting. The cost can range from 2 to 5 percent of the process cost. In
the WBS models, this category is considered to account for administrative costs that the
purchaser incurs in the course of procurement. These costs are distinct from the construction
management fee, which is included as a separate indirect cost (Section D.17). The WBS models
use a default value of 2 percent. Users can change this assumption on the indirect assumptions
sheet of each model.

D.15 Sales Tax

Water treatment plant projects may be exempt from the sales tax, particularly those constructed
with public funds. The default value in the WBS models is 0 percent, which reflects the status of
taxes in social cost analysis. Taxes are considered a transfer payment and not an actual social
cost, which is based on the concept of opportunity cost. Transfer payments are not included in
social cost analysis, so a default value of 0 percent is appropriate for social cost analysis. The
WBS models include a sales tax line item among indirect costs because the models can also be
used for private cost analysis (e.g., for a specific utility), which includes transfer payments.
Users can enter a sales tax percentage on the indirect assumptions sheet of each model in cases
where consideration of transfer payments is appropriate.

D.16 Financing During Construction

Engineering cost estimates include interest for financing of the project. Drinking water systems
can obtain financing through various sources including Drinking Water State Revolving Funds
(DWSRF), public-sector financing, private-sector borrowing or equity instruments. Exhibit D-8
shows interest rates for drinking water treatment projects derived from the EPA 2006
Community Water Systems Survey. The default value in the WBS models is 5 percent, which is
toward the higher end of the range of financing costs for public and private systems, and
implicitly assumes 1 year of financing during construction. For small systems with design flow
less than 1 MGD, the models assume 0 percent financing during construction, implicitly
accounting for the very short construction time required for these systems. Users can change the
assumptions for both large and small systems on the indirect assumptions sheet of each model.

110
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit D-8. Average Interest Rates for Capital Funds

System Ownership Type Average Interest Rate Range of Average Interest
Lender
(All System Sizes) Rates
Public DWSRF 2.3 1.0 – 3.5
Public Other Public Sector 3.5 0.5 – 4.4
Public Private Sector 4.6 4.2 – 5.0
Public Other 3.9 0.0 – 4.9
Private DWSRF 5.6 0.8 – 6.2
Private Other Public Sector 4.4 3.1 – 5.5
Private Private Sector 6.5 4.3 – 7.7
Private Other 5.9 0.0 – 10.0
All Systems DWSRF 2.6 1.0 – 4.3
All Systems Other Public Sector 3.8 1.9 – 4.5
All Systems Private Sector 5.2 4.3 – 5.5
All Systems Other 4.3 0.0 – 10.0
All Systems All Lenders Data not available 0.0 – 10.0
DWSRF = Drinking Water State Revolving Funds

D.17 Construction Management and General Contractor Overhead

As discussed in Section D.4.2, the component costs in the WBS cost database include the cost of
installation, including O&P for the installing contractor. However, the installation cost does not
cover the cost of insurance, performance bonds, job supervision or other costs associated with
the general contractor. The WBS models account for these costs by combining costs and fees for
the following items:

• Builder’s risk insurance

• Performance bonds
• Construction management.

Builder’s risk insurance is casualty insurance for the project during construction and may cover
various risks, such as vandalism, fire, theft or natural disasters. According to RSMeans (2009c),
a national average rate is 0.34 percent of the project cost. EPA adopted this assumption for the
WBS models. Users can adjust this rate on the indirect assumptions sheet of each model.

Performance bonds compensate the owner for losses due to contractor failure to complete work
according to specifications. RSMeans (2006) estimates the costs based on the total direct cost of
the project, as described in Exhibit D-9.

111
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit D-9. Cost of Performance Bonds

Project Direct Cost Range Performance Bond Cost

<$100,000 2.5%
$100,000 to $500,000 $2,500 plus 1.5% of the amount over $100,000
$500,000 to $2,500,000 $8,500 plus 1.0% of the amount over $500,000
$2,500,000 to $5,000,000 $28,500 plus 0.75% of the amount over $2,500,000
$5,000,000 to $7,500,000 $47,250 plus 0.70% of the amount over $5,000,000
>$7,500,000 $64,750 plus 0.60% of the amount over $7,500,000
Source: RSMeans (2006), reference table R013113-80.

The construction management fee covers the cost of job supervision, an on-site office, main
office overhead and profit. Various sources provide individual estimates for these items, but the
WBS models roll them into a construction management fee to reflect a cost structure that the
owner might see. RSMeans (2009c) provides a table of typical construction management rates
for jobs of various sizes. EPA adapted those rates to develop those shown in Exhibit D-10.

Exhibit D-10. Construction Management Fees

Project Direct Cost Range Construction Management Fee

<$100,000 10%
$100,000 to $250,000 9%
$250,000 to $1,000,000 6%
$1,000,000 to $5,000,000 5%
$5,000,000 to $10,000,000 4%
>$10,000,000 3.2%a
a. The reference quotes a fee range of 2.5% to 4% for a $50 million project. The WBS models assume an intermediate rate for
projects over $10 million.
Source: RSMeans (2009c), division 01 11 31.20.

The indirect line item for construction management and general contractor overhead sums all of
these costs. The costs can be omitted individually on the indirect assumptions sheet of each
model, either by an assumption that directly controls inclusion or exclusion or by setting the
appropriate percentage to zero. For example, in the case of small systems that use pre-engineered
package treatment plants, the models exclude the construction management fee portion by
default and include only the performance bond and builder’s risk insurance. Because package
plants typically are skid-mounted and assembled offsite, they typically do not require a general
contractor to supervise their installation. Instead, their installation is managed by a single entity,
often the vendor that supplied the package.

D.18 City Index

This indirect cost accounts for city-specific and regional variability in materials and construction
costs. The city index factor included in the WBS models is expressed as a decimal number,
assuming a national average of 1.0. The default value in the WBS models is set to the national
average of 1.0, which is appropriate for estimating national compliance costs. Users wishing to
adjust estimated costs to be more reflective of potential costs in specific geographic locations can

112
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

change the city index value on the output sheet. For example, to estimate costs for a city where
construction costs are 90 percent of the national average, the user would change the city index to
0.9. One source for region- or location-specific adjustment factors is RSMeans (2013), which
publishes average construction cost indices for various three-digit zip code locations.

D.19 List of Abbreviations and Symbols in this Appendix

DWSRF Drinking Water State Revolving Funds
EPA U.S. Environmental Protection Agency
ENR Engineering News-Record
ft2 square feet
I&C instrumentation and control
KW kilowatt
LTL less than a load
MGD million gallons per day
O&P overhead and profit
WBS work breakdown structure

D.20 References
Association for the Advancement of Cost Engineering International (AACEI). 1996.
Certification Study Guide. Morgantown, West Virginia.

Allis, William A. 2005. Personal communication with L. Petruzzi, SAIC. April.

Air Force Combat Climatology Center (AFCCC). 2000. Engineering Weather Data: 2000
International Edition. Published by the National Climatic Data Center.

American Water Works Corporation (AWWC). 1999. Preliminary Cost Estimate Summary:
Alton Water Treatment Plant, H&S no. 1862. 7 January.

AWWC. 2001a. Seymour, Indiana Process Cost Estimate.

AWWC. 2001b. St. Joseph, Missouri Cost Estimate.

Burger, Riaan. 2003. “Contingency, Quantifying the Uncertainty.” Cost Engineering 45, no. 8. 8
August.

California Department of Transportation (CADOT). 2001. Earth Pressure Theory and

Application.

Construction Industry Institute (CII). 2001. Benchmarking & Metrics Analysis Results. Austin,
Texas. May.

Fairfax Water Authority (FWA). 2003. New Construction Works Brochures. Virginia.

GeoEconomics Associates Incorporated. 2002. An Economic Analysis of Water Services. Chapter

5.

113
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Krynine, D.P. and W.R. Judd. 1957. Principles of Engineering Geology and Geotechnics.
McGraw-Hill. New York.

Minnesota Natural Resources Conservation Services (MN NRSC). 2003. General Guide for
Estimating Moist Soil Density. 10 May.

National Renewable Energy Laboratory (NREL). 2003. Gas-Fired Distributed Energy Resource
Technology Characterizations. U.S. Department of Energy.

RSMeans. 2006. Facilities Construction Cost Data. 21st Annual Edition. Kingston,
Massachusetts: RSMeans Company.

RSMeans. 2009a. Assemblies Cost Data. 34th Annual Edition. Kingston, Massachusetts:
RSMeans Company.

RSMeans. 2009b. Building Construction Cost Data. 67th Annual Edition. Kingston,
Massachusetts: RSMeans Company.

RSMeans. 2009c. Facilities Construction Cost Data. 24th Annual Edition. Kingston,
Massachusetts: RSMeans Company.

RSMeans. 2013. Facilities Construction Cost Data 2014. 29th Annual Edition. Norwell,
Massachusetts: Reed Construction Data LLC.

Scoutten, Inc. 2009. Opinion of Probable Cost for Town Of Buckeye Water And Wastewater
Infrastructure and Water Resources Improvements Associated with 2009 Development Fees.
Revised 11 May. Online at http://www.buckeyeaz.gov/DocumentView.aspx?DID=640

Stone and Webster. 2001. Brayton Point Station Permit Renewal Application. Five Volumes.

Tampa Bay Water. 2006. West Pasco Infrastructure Project, Maytum WTP Modifications,
Project No. 05903. Memorandum from Kenneth R. Herd, Director of Operations and Facilities,
to Jerry L. Maxwell, General Manager. 1 December.

United States Army Corp of Engineers (U.S. ACOE). 1992. Revision of Thrust Block Criteria in
TM 5-813-5/AFM 88-10 VOL. 5 Appendix C. Publication Number: ETL 1110-3-446. 20 August.

U.S. Environmental Protection Agency (U.S. EPA). 2004. Capital Costs of Arsenic Removal
Technologies, U.S. EPA Arsenic Removal Demonstration Project, Round 1. EPA-600-R-04-201.
Cincinnati, OH: U.S. EPA, National Risk Management Research Laboratory.

114
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Appendix E. General Assumptions for Operating and

Maintenance Costs
E.1 Introduction
The work breakdown structure (WBS) models calculate operating and maintenance (O&M) costs
independently for each treatment technology. Nevertheless, there are several general assumptions
and estimation functions that are common to the O&M estimates across the treatment models.
This appendix describes those assumptions and functions. Any O&M cost element that is
technology-specific is included in the chapter describing that technology in the main document.

The O&M costs estimated in the WBS models primarily include annual expenses for:

• Labor to operate and maintain the new treatment equipment

• Chemicals required by the treatment
• Materials needed to carry out maintenance (including small tools)
• Energy.

Costs for commercial liability insurance, inspection fees, domestic waste disposal, property
insurance and other miscellaneous expenditures that are not directly related to the operation of
the technology are included in the WBS models by applying a miscellaneous allowance to the
total annual O&M costs. This calculation uses the same miscellaneous allowance percentage that
is applied to capital costs as an indirect line item (see Appendix D). Users can change this
percentage on the indirect assumptions sheet of each model.

The WBS models calculate O&M costs based on the inputs provided by the user on the input
sheet and values specified on the O&M assumptions sheet. These inputs include system size, raw
and finished water quality parameters and other factors that affect operation requirements such as
an option in the activated alumina model to regenerate media or operate on a throw away basis.
The design equations and assumptions incorporated in the O&M sheets are described below.

Despite provisions for user inputs, there are several factors that can affect site-specific O&M
costs in ways that are not readily reflected in the WBS outputs. These include:

• Operator expertise
• Equipment quality, design, installation and degree of automation
• Environmental conditions (e.g., changes in raw water quality over time).

Some O&M costs components, such as energy for pumping water or chemicals for treatment, are
well defined and readily estimated using an engineering cost approach. Other O&M cost drivers,
however, depend on multiple factors that are difficult to quantify and, therefore, represent a
challenge for estimating costs. For example, the required level of effort to operate or maintain a
technology depends on the level of complexity and sophistication of the installed technology, the
size of the treatment system, the professional level or education and training of the operator and
state and local regulations for process staffing.

115
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

To complicate matters further, there are trade-offs between system capital costs and O&M costs.
Higher cost equipment may require less intensive maintenance or less hands-on operation. For
example, using mixers and tanks to prepare brine solution for regenerating an anion exchanger
might reduce equipment costs compared to using salt saturators. However, salt saturators require
less labor to use and potentially reduce the need for a salt storage facility. Also high quality,
highly automated systems can significantly reduce labor requirements, but increase capital costs.
The U.S. Environmental Protection Agency (EPA) included some adjustments to O&M costs in
the WBS models to account for some types of savings, which are described below.

E.2 Labor Costs

The WBS models calculate the annual hours of O&M labor in the following categories:

• Operator labor for operation and maintenance of process equipment

• Operator labor for building maintenance
• Managerial and clerical labor.

The WBS model labor hour estimates are intended to be incremental. That is, they only include
labor associated with the new treatment system components.

E.2.1 Operator Labor for Operation and Maintenance of Process Equipment

System operation includes the following primary tasks:

• Collecting data from process instruments and recording system operating parameters
• Preventative maintenance and calibration of process instruments
• Verifying the proper operation of pumps, valves and other equipment and controlling the
treatment process by adjusting this equipment
• Preventative maintenance of pumps, valves and other equipment
• Inspection and maintenance of chemical supplies
• Visual inspection of the treatment facility and system components
• Other, technology-specific tasks (e.g., managing regeneration, backwash or media
replacement).

Labor required for these tasks is sensitive to the level of system automation. As discussed in
Section 2.3 and Appendix A, the user has the option to choose from three levels of automation:
manual, semi-automated and fully automated. The assumptions about labor required for each
task vary depending on the level of automation selected, as shown in Exhibit E-1 and discussed
below. Users can change these assumptions, if desired, on the O&M assumptions sheet of each
model.

EPA compared model results using these assumptions with annual labor hours reported for 12
different water treatment facilities. Most of the resulting model estimates were within +50
percent to -30 percent of the annual labor hours reported for the sample facilities.

116
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Exhibit E-1. Operator Labor Assumptions for Three Levels of Automation

Task Manual Semi-automated Automated
Record system operating parameters from process 5 minutes per day
5 minutes per day 5 minutes per day
instruments (includes routine sampling) per instrument
10 minutes per 10 minutes per
Preventative maintenance and calibration of process 10 minutes per month
month per month per
instruments per instrument
instrument instrument
5 minutes per day 5 minutes per day
Verify and adjust pump operating parameters None
per pump per pump
30.25 hours per 30.25 hours per 30.25 hours per year
Preventative maintenance of pumps
year per pump year per pump per pump
5 minutes per 5 minutes per
Verify and adjust valve positions None
week per valve week per valve
5 minutes per year 5 minutes per year 5 minutes per year per
Preventative maintenance and inspection of valves
per valve per valve valve
1 minute per day 1 minute per day 1 minute per day per
Visual inspection of facility per 100 square per 100 square 100 square feet of
feet of facility feet of facility facility
120 minutes per 120 minutes per
120 minutes per month
month per month per
Inspect and maintain chemical supplies per chemical supply
chemical supply chemical supply
tank
tank tank

Collecting Data from Process Instruments and Recording System Operating

Parameters
For manual systems, the models assume 5 minutes per day per instrument associated with day-
to-day operation of the treatment process (e.g., flow meter, head loss sensor). Instruments
associated with intermittent processes (e.g., backwash flow meters) are not included in this
estimate, because observation of these instruments is included the operator labor associated with
managing the intermittent process. In semi-automated and automated systems, the control system
handles the task of collecting information from the various process instruments, so operator labor
is reduced to 5 minutes per day to keep a record of operating parameters.

Preventative Maintenance and Calibration of Process Instruments

Regardless of the level of automation, the models assume 5 minutes per month for each
instrument, including those associated with intermittent processes. While some instruments (e.g.,
chlorine residual analyzers) may require calibration more frequently than monthly, others (e.g.,
head loss sensors) will require limited, less frequent maintenance. Therefore, the models use 10
minutes per month as an average across the various types of instruments.

Verify and Adjust Pump Operating Parameters

For manual and semi-automated systems, the models assume 5 minutes per day per pump,
including metering pumps associated with continuous chemical feed processes. Pumps
associated with intermittent processes (e.g., backwash pumps) are not included in this estimate,

117
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

because operation of these pumps is included the operator labor associated with managing the
intermittent process. In automated systems, the control system handles this task, so no additional
operator labor is required.

Preventative Maintenance of Pumps

Regardless of level of automation, the models assume 30.25 hours per year for large, frequently
operated pumps (e.g., booster pumps). This estimate does not include small chemical metering
pumps, but does include backwash pumps when these are operated more frequently than weekly.
The estimate of 30.25 hours per year is based on a list of recommended pump maintenance
activities from a vendor, combined with an independent engineering estimate of hours required
for each activity, as shown in Exhibit E-2.

Exhibit E-2. Pump Maintenance Activities

Estimated
Task Interval Estimated Hours/Year
Minutes/Task
Check bearing temperature Monthly 5 1
Changing lubricant/ adjusting power level Monthly 30 6
Disassemble for inspection, reassemble Monthly 60 12
Check oil Quarterly 10 0.67
Check lubricated bearings for saponification Quarterly 10 0.67
Removal of bearings and replace, reassemble Quarterly 60 4
Check packing and replace if necessary, reassemble 6 months 60 2
Vibration readings 6 months 10 0.33
Remove casing and inspect pump Annual 120 2
If parts are worn, replace Annual varies covered by pump materials
percentage and pump
useful life
Clean deposits and/ or scaling Annual 60 1
Clean out stuffing box piping Annual 30 0.5
Measure and record suction and discharge pipe head Annual 5 0.08
Total All Tasks Annual 1,815 30.25

Verify and Adjust Valve Positions

For manual and semi-automated systems, the models assume 5 minutes per week per valve on
the main process line. In automated systems, the control system handles this task, so no
additional operator labor is required.

Preventative Maintenance and Inspection of Valves

Regardless of level of automation, the models assume 5 minutes per year per valve on the main
process line.

118
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Visual Inspection of Facility

Regardless of level of automation, the models assume 1 minute per week per 100 square feet (ft2)
of treatment system floor plan to conduct visual inspection of the overall process. This daily
inspection is in addition to inspection conducted as part of routine operation and maintenance of
major operational components (instruments, pumps and valves), as discussed above.

Inspect and Maintain Chemical Supplies

The models assume that chemical supplies, whether they are associated with continuous addition
or intermittent use, require additional attention beyond that included in daily visual inspection. In
particular, they also require labor associated with receiving chemical shipments. Regardless of
level of automation, the models assume 120 minutes per month for each chemical storage tank.
Although counted on the basis of the number of tanks, this estimate is intended to cover all
components associated with the chemical supply system (e.g., checking pipes and valves for
leaks and inspecting and maintaining small metering pumps).

Technology-Specific Tasks
Many of the technologies include activities in addition to day-to-day operation that may require
operator attention, depending on the level of automation (e.g., intermittent regeneration,
backwash or media replacement). Where this is the case, the technology chapters in the main
document describe the specific assumptions required to calculate operator labor.

E.2.2 Labor for Building Maintenance

The WBS models include a building maintenance cost based on the building area (i.e., using a
unit cost in dollars per square foot per year). EPA developed this cost based on two sources:
Whitestone Research (2009) and RSMeans (2013). Specifically, EPA selected a list of
appropriate building maintenance repair and repair tasks from those listed in the two sources.
The selected tasks include those associated with preventative maintenance, small repairs and
major repairs. EPA estimated a frequency for each task by averaging the frequency
recommended in each of the two sources. Exhibit E-3 identifies the tasks included in the
maintenance and repair buildup. The models include maintenance and repair tasks for heating
and cooling systems only for buildings with the relevant systems. To avoid double-counting, the
task list does include tasks in the following categories:

• Maintenance of treatment system components that are already explicitly considered in the
models’ maintenance labor and materials costs (e.g., pumps, valves, instruments)
• Full replacement of items that are explicitly given a useful life in the models (e.g., piping,
heating and cooling systems)
• Repair tasks with a lower frequency than the useful life assumed in the models for the
related WBS line item (e.g., skylight replacement has a recommended frequency of 40
years, whereas the models assume a useful life for the entire building of 37 to 40 years).

For the buildup, EPA assumed a baseline building area of 4,000 ft2 and building components
corresponding to a medium-quality building (see the assumptions in Appendix B). EPA
estimated costs for each task using data from RSMeans (2013) and assuming that preventative
maintenance and minor repairs would be conducted using in-house labor, while major repairs
would be conducted using outside contractors. These costs include both labor and materials.

119
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Because repair needs do not follow a strict schedule, EPA annualized costs with no discount
rate—that is, a $100 task with a typical frequency of 5 years is assigned an annual cost of $20.

Labor accounts for most of the cost for the maintenance and repair tasks. Further, the Building
Cost Index and Construction Cost Index, the only two cost indices in the WBS cost database that
combine labor and material costs, do not include the costs of materials that are likely to be used
in building maintenance. The WBS cost database therefore uses the Employment Cost Index to
escalate the building maintenance costs to current year dollars.

Exhibit E-3. Building Maintenance and Repair Tasks

• Minor repairs and refinishing for concrete floors
• Repairs and waterproofing for exterior concrete block walls
• Repairs and refinishing for doors
• Roofing debris removal and inspections
• Minor repairs and replacement for roofing membranes and flashing
• Repairs to skylights
• Repairs and refinishing for interior concrete block walls
• Repairs and refinishing for drywall
• Office painting
• Vinyl tile flooring replacement
• Repairs, refinishing and replacement for acoustic tile ceilings
• Preventive maintenance, repairs and replacement for lavatories and lavatory fixtures
• Water heater preventive maintenance, cleaning and servicing, overhauls and replacement
• Repairs to pipe joints and fittings
• Cleaning of drains
• Maintenance, repair and replacement of gutters
• Repair and replacement of fans
• Inspection and replacement of sprinkler systems
• Maintenance, inspection, repair and replacement of electrical systems including switchgear, receptacles, wiring devices,
voice/data outlets and structure ground
• Replacement of lamps, ballasts and lighting fixtures
• Standby generator maintenance and inspection
• Preventive maintenance of computers

E.2.3 Managerial and Administrative Labor

The models contain an assumption that managerial and administrative support levels for a new
treatment plant are equal to 10 percent of the total operator hours for system operation and
maintenance. This estimate is not intended to represent total administrative and managerial time
at a drinking water system, because total time includes many tasks unrelated to operating a new
treatment train. It only represents incremental time needed to provide administrative support for
the new treatment plant, e.g., processing supply orders. It also does not include labor time
associated with recordkeeping and reporting burden estimates that EPA must estimate and report
independently to comply with Paperwork Reduction Act requirements. Users can change the 10

120
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

percent assumption for either or both of managerial and administrative labor on the O&M
assumptions sheet of each model.

E.2.4 Labor Unit Costs

To estimate the cost of the labor calculated in each of the categories above, the models multiply
labor hours by unit costs from the WBS cost database. These unit costs reflect average loaded
wage rates for applicable labor categories (i.e. technical, managerial and administrative) and vary
across system size. The WBS cost database uses the Employment Cost Index to escalate the rates
to current year dollars. Users can change the wage rates, if desired, in the data extracted from the
WBS cost database.

E.3 Chemicals
Each of the WBS models calculates annual chemical usage (in pounds or gallons per year) on a
technology-specific basis. The technology chapters in the main document describe these
calculations. In some models, these calculations also reflect the selected option for regeneration
or disposal of spent chemicals. Annual chemical costs equal the product of the annual chemical
requirements and the unit chemical costs in the WBS cost database.

E.4 Materials
The WBS models calculate the annual cost of materials in the following categories:

• Materials for maintenance of booster or influent pumps

• Materials for maintenance and operation of other, technology-specific equipment
• Replacement of technology-specific equipment that occurs on an annual basis
• Materials for building maintenance.

Pumps are operated continuously (or nearly continuously) and require preventive and routine
maintenance. Pumps are common to all the technologies. Each of the models assumes the annual
cost of materials for pumps is equal to 1 percent of their pre-installation capital cost to account
for consumable supplies and small parts requiring frequent replacement. This assumption is
based on input from the technology experts who reviewed the WBS models and commented that
the initial assumption of 5 percent was too high. Users can change this assumption on the O&M
assumptions sheet of each model. Although accidents or improper operation can result in a need
for major repairs that increase maintenance materials costs beyond 1 percent, the models do not
include these types of costs.

Some of the technologies include other equipment that may require significant maintenance (e.g.,
the blowers in the packed tower aeration and multi-stage bubble aeration technologies). The
models for these technologies include annual costs for maintenance materials. The technology
chapters in this document describe the specific calculations. In general, these calculations are
based a percentage of the pre-installation capital cost of the equipment.

Some of the technologies also include equipment components (e.g., membrane filter cartridges)
that require frequent replacement. Rather than treat these components as frequently replaced
capital items, the models handle the replacement costs in the O&M sheet. The replacement cost
calculations are based on assumptions about replacement frequency and unit costs from the WBS

121
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

component cost database. The specific calculations are in the technology sections of the main
document.

The WBS models compute a cost for building maintenance that combines labor and materials.
The cost is discussed in Section E.2.2.

E.5 Energy
All of the WBS models calculate the annual cost of energy in the following categories:

• Energy for operation of booster or influent pumps

• Energy for operation of other, technology-specific items of equipment
• Energy for lighting
• Energy for ventilation
• Energy for cooling
• Energy for heating.

E.5.1 Energy for Pumps and Other Equipment

Booster or influent pumps are equipment common to all the technologies. Because these pumps
are operated continuously (or nearly continuously), they can represent significant energy
consumption. Therefore, each of the WBS models calculates pump operating horsepower based
on average flow, pump head and pump efficiency. The models then convert this operating
horsepower to megawatt-hours/year assuming continuous operation. To calculate annual cost, the
models then multiply the annual power requirement by the unit cost for electricity contained in
the WBS component cost database.

Some of the technologies include other equipment that consumes significant quantities of energy
(e.g., blowers, backwash pumps, mixers). For those technologies, the model also calculates the
energy for such equipment explicitly. The technology chapters in the main document describe
the specific energy calculations. In general, those calculations are similar to the energy
calculation for pumps.

E.5.2 Energy for Lighting

The models calculate annual lighting requirements based on the building square footage estimate
and the quality level of the building (see Appendix B). The building capital costs in the WBS
cost database include the cost of light fixtures for the following light requirements:

• Sheds and low quality buildings, 1 watt/hour/ft2 of building area

• Mid quality buildings, 2 watts/hour/ft2 of building area
• High quality buildings, 4 watts/hour/ft2 of building area.

Multiplying the appropriate light requirement by 8.8 results in an energy usage rate in kilowatt
hours per ft2 per year. This conversion is based on operation of the lights 24 hours per day, 365
days per year. EPA evaluated these assumptions by calculating the granular activated carbon
contactor, pipe gallery and furnace area lighting requirements at the Richard Miller Water
Treatment Plant in Cincinnati, Ohio. For this facility, the lighting requirements are 1.5, 1.0 and
0.8 watts per hour per ft2 for the contactor, pipe gallery and regeneration areas, respectively, with

122
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

a weighted average of 1.0 watt per hour per ft2, which is at the low end of the range that EPA
uses in the models. Technologies with different types of process equipment that require more
frequent access may require more lighting. Users can change the lighting requirement for each
level of building quality on the O&M assumptions sheet of each model.

Because many systems are not lit during times an operator is not present, the models reduce
lighting energy requirements when a full-time operator presence is not required (possible for
small systems for many technologies) using the following factor (with a maximum of 1 to
account for large systems that might require more than one full-time operator):

Operator hours per year/ (24*365).

E.5.3 Energy for Ventilation

The models calculate ventilation requirements based on the assumptions shown in Exhibit E-4.
The technology experts who reviewed the assumptions for the WBS models confirmed the
reasonableness of these assumptions, although one expert commented that the air change rate for
pumps could be lower for systems in a northern climate. The WBS models continue to use the
value shown, however, because it is believed to be more reasonable for a national average
estimate and results in a more conservative (i.e., higher) estimate of ventilation energy
consumption. All of the models use these same assumptions with the exception of chlorination,
which has special ventilation requirements as described in that technology section. Users can
change the ventilation assumptions, if desired, on the O&M assumptions sheet of each WBS
technology model.

Exhibit E-4. Assumptions for Calculating Ventilation Requirements

Variable Value used
Ventilation air change rate for contactor areas 3 air changes/hour
Ventilation air change rate for pump areas 20 air changes/hour
Ventilation air change rate for chemical storage areas 2 air changes/hour
Ventilation air change rate for offices 2 air changes/hour
Pressure drop across ventilation fans 0.25 pounds/ ft2
Number of days with mechanical ventilation for small systems (less than 1 MGD) 90 days/year
Number of days with mechanical ventilation for medium systems (1 to 10 MGD) 120 days/year
Number of days with mechanical ventilation for large systems (greater than 10 MGD) 185 days/year
Building height 20 feet
MGD = million gallons per day

The models first use the air change rate assumptions to calculate an overall weighted average air
change rate for each building based on the equipment present in that building. The models then
use this weighted average air change rate for each building in the following formula:

Ventilation energy (MWh/yr) = DAYS × 24 × 0.746 × Pdrop × FP × H × Achanges / 33,000,000

where:

123
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

DAYS = days per year with mechanical ventilation

Pdrop = pressure drop across ventilation fans (pounds/ft2)
FP = building footprint (ft2)
H = building height (feet)
Achanges = weighted average air change rate for the building (air changes/hour)

E.5.4 Energy for Heating and Cooling

The models calculate heating and cooling requirements based on the assumptions shown in
Exhibit E-5. Users can change these assumptions, if desired, on the O&M assumptions sheet of
each WBS technology model. These assumptions are described in greater detail below.

R-values are a measure of the effective resistance to heat flow of an insulating barrier such as a
building envelope. The R-values assumed in the models (13 for walls, 38 for ceilings) are based
on the use of standard building materials. The user can change these values to reflect higher
efficiency construction materials. In doing so, however, the user should also examine the unit
building costs extracted from the WBS cost database (see Appendix B) to determine if they are
consistent with the use of such construction materials.

Exhibit E-5. Assumptions for Calculating Heating and Cooling Requirements

Variable Value used
R-value for walls 13 hour - ft2 - oF /BTU
R-value for ceilings 38 hour - ft2 - oF /BTU
Annual heating degree days 4,923 degree days
Annual cooling degree days 1,697 degree days
Heating ventilation/infiltration load 168,679 BTU/cfm
Cooling ventilation/infiltration load 51,771 BTU/cfm
Electric resistance heating efficiency 98%
Heat pump heating coefficient of performance 3.2
Natural gas non-condensing furnace efficiency 80%
Natural gas condensing furnace efficiency 95%
Diesel non-condensing furnace efficiency 78%
Diesel condensing furnace efficiency 85%
Air conditioning energy efficiency ratio 11 Whr/BTU
Heat pump cooling energy efficiency ratio 10.1 Whr/BTU
Maximum capacity for heat pump heating 200 thousand BTU per hour
Maximum capacity for other heating options 6,148 thousand BTU per hour
Maximum capacity for heat pump cooling 50 tons
Maximum capacity for other cooling options 113.32 tons
BTU = British thermal unit; cfm = cubic feet per minute; Whr = watt hour

124
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

The next four values in the exhibit (annual heating and cooling degree days and heating and
cooling ventilation/infiltration loads) are climate-related. EPA derived these values from data in
the Air Force Combat Climatology Center Engineering Weather Data Version 1.0 (U.S. Air
Force, 2000). Specifically, EPA selected climate data for 21 cities distributed geographically
throughout the United States and calculated total annual heating and cooling losses. The values
used in the WBS models are those for the city (St. Louis) that represents the median total annual
heating and cooling loss from among the 21 cities. Therefore, the values used are intended to
represent a climate that produces a national median total heating and cooling requirement. The
user can change these values to represent a specific different climate. In doing so, however, the
user should select values for the heating and cooling measures, respectively, that are consistent
with one another (i.e., reflective of a realistic climate).

The remaining values in the exhibit are related to the efficiency and performance of heating and
air conditioning equipment. These values are based on data from the following sources:

• 10 CFR 430.32
• Canadian Office of Energy Efficiency (2009)
• ACEEE (2012).

The user can modify these values, as desired, to reflect the use of more or less efficient
equipment.

The WBS models use the assumptions in Exhibit E-5, along with estimated building
dimensions, to calculate total annual heating and cooling losses. The models consider both
conductance losses and ventilation/infiltration losses. The models calculate conductance losses
for each building using the following formulae:

Conductance heating loss = 4 × S × H × HDD × 24 / Rwall + FP × HDD × 24 / Rceiling

Conductance cooling loss = 4 × S × H × CDD × 24 / Rwall + FP × CDD × 24 / Rceiling

where:
S = length of building side in feet (assumed to equal the square root of the building
footprint)
H = building height (feet)
HDD = annual heating degree-days
CDD = annual cooling degree-days
FP = building footprint (ft2)
Rwall = R-value for walls
Rceiling = R-value for ceiling.

The equations above represent the total heat transfer in British thermal units (BTU)/year through
all four walls and the ceiling of each building. The models assume heat transfer through the
building floor is negligible.

To calculate ventilation and infiltration losses, the models first calculate the air exchange rate in
cubic feet per minute (cfm) for each building using the following formula:

125
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

Air exchange rate (cfm) = FP × H × Achanges / 60

where:
H = building height (feet)
FP = building footprint (ft2)
Achanges = weighted average air change rate for the building (air changes/hour, as
described above in the section on ventilation)

Note that, unlike the calculation for ventilation energy use, this calculation does not incorporate
assumptions about the frequency of mechanical ventilation. This is because heating and cooling
losses occur regardless of whether ventilation is achieved by mechanical or natural means.

The models then apply the air exchange rate calculated above to determine ventilation and
infiltration losses (in BTU/year) for each building using the following formulae:

Heating ventilation and infiltration heat loss = CFM × HVIload

Cooling ventilation and infiltration heat loss = CFM × CVIload

where:
CFM = air exchange rate (cfm, as calculated above)
HVIload = heating ventilation/infiltration load (in BTU/cfm)
CVIload = cooling ventilation/infiltration load (in BTU/cfm)

The models then sum conductance losses and ventilation/infiltration losses to determine total
annual heating and cooling requirements for each building. For cooling, the models add cooling
required to compensate for the waste heat generated by pumps (and other technology-specific
mechanical equipment).

The models then calculate heating and cooling energy consumption for each of several options
using these requirements, BTU values for the appropriate fuel (i.e., electricity, natural gas or oil)
and the efficiency factors shown in Exhibit E-5. For heating, the options are electric resistance
heating, electric heat pump, natural gas condensing or non-condensing furnace and diesel
condensing or non-condensing furnace. For cooling, the options are conventional air
conditioning and electric heat pump. As indicated by the final two assumptions in Exhibit E-5,
some of these options are not applicable for systems requiring large heating or cooling capacity
(i.e., large systems with large buildings). Specifically, electric resistance heating and heat pumps
are not useable for heating beyond a maximum capacity of 240,000 BTU/hour and heat pumps
are not useable for cooling beyond this same maximum capacity. When total annual heating or
cooling losses are greater than these maximum capacities, the models do not include these
options among those displayed on the output sheet.

The models determine whether to include heating and cooling costs (both capital and O&M)
based on building size, system design flow and user input for component level (see Section 2.3),
as shown in Exhibit E-6. Users can change these assumptions on the indirect assumptions sheet
of each model. When heating and/or cooling are included, the models choose among the heating
and cooling system options based on the total annualized cost of each option (annual energy cost,
plus capital cost of the system annualized as discussed in Section 2.4). The models select the

126
 WBS-Based Cost Model for Reverse Osmosis/Nanofiltration Drinking Water Treatment

option with the lowest annualized cost for inclusion in the system capital costs and add the
corresponding annual energy cost to O&M costs.

Exhibit E-6. WBS Model Assumptions Regarding Inclusion of Heating and Cooling
Building Square Component Cost System Design Flow: System Design Flow: System Design Flow:
Footage Level Selected Less than 1 MGD 1 to 10 MGD 10 MGD or greater
500 or greater Low Neither Heating Only Heating and Cooling
500 or greater Medium Heating Only Heating and Cooling Heating and Cooling
500 or greater High Heating and Cooling Heating and Cooling Heating and Cooling
Less than 500 Low or Medium Neither Neither Heating Only
Less than 500 High Heating Only Heating Only Heating Only
MGD = million gallons per day

E.5.5 Energy Unit Costs

To estimate the cost of the energy consumption calculated in each of the categories above, the
models use unit costs from the WBS component cost database. These unit costs represent
national averages for each fuel (electricity, natural gas and diesel) obtained from the U.S.
Department of Energy’s Energy Information Administration. Because energy costs are highly
variable, users can change these energy unit costs, if desired, in the data extracted from the WBS
cost database.

E.6 List of Abbreviations and Symbols in this Appendix

BTU British thermal unit
cfm cubic feet per minute
EPA U.S. Environmental Protection Agency
ft2 square feet
O&M operating and maintenance
WBS work breakdown structure

E.7 References
American Council for an Energy Efficient Economy (ACEEE). 2012. Heating. Last updated
December. Online at http://www.aceee.org/consumerguide/heating.htm.

Canadian Office of Energy Efficiency. 2009. Heating with Gas – Comparing Annual Heating
Costs.

RSMeans. 2013. Facilities Maintenance and Repair Cost Data 2014. 21st Annual Edition.
Norwell, Massachusetts: Reed Construction Data.

U.S. Air Force. 2000. Air Force Combat Climatology Center Engineering Weather Data.
Version 1.0.

Whitestone Research. 2009. The Whitestone Facility Maintenance and Repair Cost Reference
2009-2010. Fourteenth Annual Edition. October.

127