EPA 600/R- l4/089 July 2014 | www.epa.gov/ord
United States
Environmental Protection
Agency
Universal Industrial Sectors Integrated Solutions (U-ISIS) Model
for the Portland Cement Manufacturing Industry
.WtrCMUr ovfsum Cwipww!
Sulk Dispatch Puking MkMk Bjj Pilleiisaticn
Office of Research and Development
-------
SEPA
United States
Environmental Protection
Agency
Universal Industrial Sectors Integrated Solutions
(U-ISIS) Model for the Portland Cement
Manufacturing Industry
July 2013
Version 3.0
EPA/600/ R-14/089
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Office of Research and Development
Air Pollution Prevention and Control Division
and
Office of Air Quality Planning and Standards 1
Sector Policies and Programs Division
-------
July 2013
Contact Information
This document has been prepared by the U.S. Environmental Protection Agency's (EPA) Office of
Research and Development (ORD) and the Office of Air Quality Planning and Standards (OAQPS).
Questions related to this document should be addressed to Dr. Gurbakhash S. Bhander, ORD,
and Elineth Torres, OAQPS, Sector Policies and Programs Division, Research Triangle Park,
North Carolina 27711 (emails: (email: bhander. gurbakhash@epa. gov and torres.elineth@epa.gov).
Acknowledgements
A large model such as U-ISIS could not have been completed without the support and advice of
many individuals and organizations. We gratefully acknowledge the following contributions.
Professor Richard Newell of Duke University and Dr. Dallas Burtraw of the Resources for the
Future (RFF) engaged in many helpful discussions, which helped shape U-ISIS. Additionally, Dr.
Burtraw helped develop the formulation of cement imports. Under EPA Contract EP-D-06-118,
Mr. Brooks Depro of RTI International developed the cost functions and other economic data for
the cement industry. Also under EPA Contract EP-C-07-015, Dr. James Staudt of Andover
Technologies developed the technical information on controls and energy efficiency measures
used in the current version of the U-ISIS cement module. Dr. Jayant Sathaye and Dr. Tengfang
Xu of Lawrence Berkeley National Laboratory (LBL) provided many useful insights on energy
efficiency measures for the cement industry. The Portland Cement Association (PCA) provided
helpful comments on U-ISIS formulation, engaged in many fruitful discussions with the U-ISIS
development team, and provided some needed data. The U-ISIS cement module was peer
reviewed by a number of experts. These experts included Dr. Dallas Burtraw of RFF, Dr. Hendrik
G. van Oss of the U.S. Geological Survey, Mr. Charles Drummond, Mr. Chris Nichols, and Mr.
Sikander Khan of the U.S. Department of Energy, Mr. Hector Ybanez of Holcim Cement, Mr.
Kevin Culligan of EPA, Dr. Fereidun Feizollahi of California Air Resources Board (Economic
Studies Section), Mr. Eric Smith of EPA, and Dr. Ernst Worrell of LBL. Mr. Charles Fulcher and
Mr. Tom Walton of EPA provided many quality assurance checks on the U-ISIS cement module.
Finally, Dr. Samudra Vijay (Sam Analysis Solutions LLC), Dr. Wojciech Jozewicz, Kim Egler, and
the staff at ARCADIS Inc. compiled the many comments received from peer reviewers and helped
with development of this documentation under EPA Contract EP-C-09-027.
-------
Table of Contents
List of Tables iii
List of Figures iv
Acronyms, Initializations, and Abbreviations v
Conversion Table - English Units to SI Units vii
Chapter 1 Introduction 1-1
1.1 The U.S. Cement Industry 1-1
1.1.1 Cement Types and Categories 1-1
1.1.2 Overview of the Cement Manufacturing Process 1-3
1.1.3 Kiln Types and Their Use 1-7
1.1.4 Portland Cement Production in the U.S 1-8
1.1.5 Imports of Portland Cement in the U.S 1-9
1.1.6 Cement Demand Centers 1-11
1.2 Emissions from the U.S. Cement Industry and Applicable Regulations 1-11
1.3 Overview of U-ISIS 1-12
1.4 References for Chapter 1 1-15
Chapter 2 U-ISIS Framework 2-1
2.1 Objective Function 2-3
2.2 Constraints 2-6
2.3 Optimization and Post Processing 2-8
2.4 U-ISIS Interface 2-8
2.4.1 Interface Features 2-9
2.4.2 Interface Data Structure 2-10
2.5 References for Chapter 2 2-12
Chapter 3 Cement Data 3-1
3.1 Data Requirements 3-1
3.2 Cement-Specific Data 3-2
3.2.1 Industry, Fuel, and Emissions 3-2
3.2.2 Control Technologies and Emission Abatement Approaches 3-12
3.2.3 Policy and Economic Parameters 3-12
3.3 References for Chapter 3 3-23
Chapter 4 Model Calibration 4-1
4.1 Calibration Methodology 4-1
4.2 Data for Calibration 4-2
-------
4.2.1 Cement Prices in USGS Districts 4-2
4.2.2 Cement Production in USGS Districts 4-3
4.2.3 Cement Imports by Import Districts 4-3
4.3 Results of Calibration 4-4
4.4 Recommendations 4-8
Appendix A
Andover Technology Partners and RTI International Memos
Appendix B
U-ISIS Mathematical Framework
n
-------
List of Tables
Table 1-1. Typical Average Heat Input by Cement Kiln Type 1-7
Table 1-2. Number of Kilns by Kiln Type in the U.S. in 2005 and 2009 1-8
Table 1-3. Largest Hydraulic Cement and Clinker Import Custom Districts in the
U.S. in 2005 1-10
Table 3-2. Portland Cement Demand in Millions of Metric Tons (2009 projections) ..3-4
Table 3-3. Bulk Shipment Costs (Cents per Metric Ton per Mile) 3-6
Table 3-4. Portland Cement and Clinker Imports in Million Metric Tons, by Major
USGS Customs District in 2010 3-7
Table 3-5. Specific Fuel Consumption and Total Exhaust Gas Flow Rate (wet) for
Various Kiln Types 3-9
Table 3-6. Estimated Uncontrolled NOx Emission Intensities for Cement Kilns 3-9
Table 3-7. Average S02 Emissions for Each Kiln Type in Each State 3-10
Table 3-8. Approximate CO2 and H2O Produced from Combustion of Fuels 3-11
Table 3-9. NOx Control Technologies for Cement Kilns 3-14
Table 3-10. S02 Control Technologies for Cement Kilns 3-17
Table 3-11. C02 Control Technologies for Cement Kilns 3-18
Table 3-12. HC1, Hg, andTHC Control Technologies for Cement Kilns 3-19
Table 3-13. Multimedia Impacts of Process Capacity Replacement on Cement Kiln
Operation1 3-19
Table 3-14. Energy Efficiency Measures for Raw Materials Preparation 3-20
Table 3-15. Energy Efficiency Clinker Making Measures 3-21
Table 3-16. Energy Efficiency Measures for Finish Grinding 3-21
Table 3-17. Energy Efficiency Plant-wide Measures 3-22
Table 4-1. Cement Prices ($/ton) for USGS Districts 4-2
Table 4-2. Cement Production (tons) for Various USGS Districts 4-3
Table 4-3. Cement Imports (except from Canada and Mexico) for Import Districts
(tons) 4-4
Table 4-4. Reported and Calculated Cement Prices in USGS Districts (2005) 4-5
Table 4-5. Reported and Calculated Cement Prices in USGS Districts (2006) 4-6
Table 4-6. Reported and Calculated Cement Prices in USGS Districts (2007) 4-7
Table 4-7. Aggregate production (reported and modeled) for 2005-2007 4-8
111
-------
List of Figures
Figure 1-1. Schematic of the Wet Cement Process 1-5
Figure 1-2. Schematic of the Dry Cement Process with Cyclone Preheater 1-6
Figure 1-3. Trends in Cement Kiln Type and Capacity in the U.S. (1995 to 2008) 1-8
Figure 1-4. Portland Cement Plant Locations 1-9
Figure 1-5. Imports of Clinker and Cement from 1998 to 2008 (PCA, 2009a) 1-11
Figure 1-6. Portland Cement Facilities and 03 NAA, PM2.5 NAA, and Class 1 Areas.. 1-13
Figure 1-7. An Integrated View of Pollution Generation Pathways, Emissions
Abatement Approaches, and Multimedia Impacts for an Industrial
Sector 1-14
Figure 2-1. Total Surplus in a Market 2-1
Figure 2-2. Modular Architecture ofU-ISIS 2-2
Figure 2-3. Stepwise Integration of the Inverse Demand Curve 2-5
Figure 2-4. Cement Modeling Framework 2-9
Figure 2-5. Interface of the U-ISIS 2-10
Figure 3-1. Modules and Associated Information Flows Utilized in the U-ISIS
Framework 3-1
Figure 3-2. Distribution of Cement Kilns in the United States as of 2009 3-3
Figure 3-3. Commercial Fuel Use Profile by U.S. Cement Industry in 2005 3-8
IV
-------
Acronyms, Initializations, and Abbreviations
ACI Activated Carbon Injection
AEO Annual Energy Outlook
APCA American Portland Cement Alliance
APPCD Air Pollution Prevention and Control Division
ASD Adjustable Speed Drives
ASTM American Society for Testing and Materials
AVC Average Variable Cost
AZ Arizona
BACT Best Available Control Technology
BAU Business As Usual
BFS Blast Furnace Slag
BLS Bureau of Labor Statistics
BTS Bureau of Transportation Statistics
CA California
CCS Carbon Capture and Sequestration
CEMBUREAU European Cement Association based in Brussels
CGC Convert to Reciprocating Grate Cooler
CKD Cement Kiln Dust
CO Colorado (in state listings only)
CO Carbon Monoxide
CO2 Carbon Dioxide
CSI Combustion System Improvement
DE Delaware
EGFW Exhaust Gas Flow Rate (wet)
EIA Energy Information Administration
EMCS Energy Management and Control System
EMD Efficient Mill Drives
EMPC Energy Management and Process Control
EPA U.S. Environmental Protection Agency
ETS Efficient Transport System
FL Florida
FOM Fixed Operation and Maintenance
GA Georgia
GAMS General Algebraic Modeling System
GDP Gross Domestic Product
GHG Greenhouse Gas
H20 Water
HAP(s) Hazardous Air Pollutant(s)
HC1 Hydrochloric Acid
HEC High-Efficiency Classifiers
HEM High Efficiency Motors
HERM High Efficiency Roller Mill
Hg Mercury
HPRP High-Pressure Roller Press
HRPG Heat Recovery for Power Generation
ID Idaho
IF Indirect Firing
v
-------
IGMBM Improved Grinding Media (Ball Mills)
IL Illinois
ISIS Industrial Sectors Integrated Solutions
LNB Low NOx Burner
MI Michigan
MMBtu Million British Thermal Units
MO Missouri
MYB Minerals Yearbook (USGS)
N/ A Not Available
NAA Nonattainment Area
NACT North American Cement Transportation
NAS National Academy of Science
NATA National Air Toxics Assessment
NC North Carolina
NEI National Emission Inventory
NESCAUM Northeast States for Coordinated Air Use Management
NESHAP National Emission Standards for Hazardous Air Pollutants
NG Natural Gas
NM New Mexico
NOx Oxides of Nitrogen (also Nitrogen Oxides)
NRC National Research Council
NSPS New Source Performance Standards
NV Nevada
O3 Ozone
O&M Operations and Maintenance
OAQPS Office of Air Quality Planning and Standards
OCAS Optimization of Compressed Air Systems
OGR Optimize Grate Cooler
OK Oklahoma
OMB U.S. Office of Management and Budget
OR Oregon
ORD Office of Research and Development
PA Pennsylvania
PCA Portland Cement Association
PCVM Process Control Vertical Mill
PM Particulate Matter
PRB Powder River Basin
PSD Prevention of Significant Deterioration
QAPP Quality Assurance Project Plan
RMB Raw Materials Blending
RMTHEC Raw Materials Transport High Energy Classifiers
ROW Rest of the World
RTO Regenerative Thermal Oxidizer
SBH Slurry Blending and Homogenization
SC South Carolina
SCR Selective Catalytic Reduction
SFC Specific Fuel Consumption
SHLR Shell Heat Loss Reduction
SNCR Selective Non-Catalytic Reduction
VI
-------
502 Sulfur Dioxide
503 Sulfur Trioxide
SPE Spatial Price Equilibrium
SR Seal Replacement
TDF Tire Derived Fuel
THC Total Hydrocarbons
TRAGIS Transportation Routing Analysis Geographic Information
System
U-ISIS Universal ISIS
USGS United States Geological Survey
UT Utah
WA Washington
WDOT Washington State Department of Transportation
WMCCC Wash Mills with Closed Circuit Classifier
WV BIT West Virginia Bituminous Coal
Conversion Table - English Units to SI Units
To Obtain
From
Multiply by
M
ft
0.3048
m2
ft2
9.29 x lO2
m3
ft3
2.83 x lO2
°C
°F
5/9 x (°F - 32)
kg
lb
0.454
J/kg
Btu/lb
1.33 x 10-4
m3/s
cfm
4.72 x 10-4
m3/s
gPm
6.31 x 10-5
J/kWh
Btu/kWh
1055.056
Mills
$
0.001
kg/m2
in. Hg
345.31
metric ton
short ton
0.907
PLEASE NOTE: for the purpose of this document, short tons will be referred to as "tons." Short ton
(ton) equals 2000 lbs. Metric tons will be referred to in this document as "metric tons." Metric ton
equals 1000 kg.
Vll
-------
Chapter 1
Introduction
In the National Academy of Science's 2004 report, "Air Quality Management in the
United States," the National Research Council (NRC) recommended to the U.S.
Environmental Protection Agency (EPA) that standard setting, planning, and control
strategy development should be based on integrated assessments that consider multiple
pollutants, and that these integrated assessments should be conducted in a
comprehensive and coordinated manner (NAS, 2004). With these recommendations,
EPA began to move toward establishing multipollutant and sector-based approaches to
manage air quality and environmental protection. The benefits of multipollutant and
sector-based analyses and approaches include the ability to identify optimum strategies
(considering feasibility, costs, and benefits across all pollutant types such as criteria,
toxics, and others), while streamlining administrative and compliance complexities and
reducing conflicting and redundant requirements.
The development of policy options for managing emissions and air quality can be made
more effective and efficient through sophisticated analyses of relevant technical and
economic factors. Such analyses are greatly enhanced by the use of an appropriate
modeling framework. Accordingly, the Universal Industrial Sectors Integrated Solutions
(U-ISIS) model has been developed at EPA (ARCADIS, 2010). Currently, the U-ISIS model
is populated with U.S. cement manufacturing data. This module has undergone external
peer review and comments have been addressed. Efforts are underway to build
representations of the U.S. pulp and paper sector and the U.S. iron and steel sector.
This document describes the framework of EPA's U-ISIS model and its application to
the U.S. cement manufacturing industry.
1.1 The U.S. Cement Industry
1.1.1 Cement Types and Categories
Cement is a finely ground powder which, when mixed with water, forms a hardening
paste of calcium silicate hydrates and calcium aluminate hydrates. Cement is used in
mortar (to bind together bricks or stones) and concrete (bulk rock-like building material
made from cement, aggregate, sand, and water). Concrete production uses the majority
of cement produced.
Portland cement and blended cement are used in concrete production, but Portland
cement is by far the most common type of cement used for concrete production. By
modifying the raw material mix and, to some degree, the temperatures utilized in
manufacturing, slight compositional variations can be achieved to produce Portland
cements with slightly different properties. In the U.S., the different varieties of Portland
cement are denoted per the American Society for Testing and Materials (ASTM)
1-1
-------
Specification C-150. The ASTM standard C-150 recognizes eight types of Portland
cement:
• Type I is for use in general construction (e.g., buildings, bridges, floors, etc.).
• Type IA is similar to Type I with the addition of an air-entraining agent.
• Type II generates less heat at a slower rate and has a moderate sulfate-attack
resistance.
• Type IIA is similar to Type II with the addition of an air-entraining agent.
• Type III is used when concrete must set and gain strength rapidly.
• Type IIIA is similar to Type III with the addition of an air-entraining agent.
• Type IV has low heat of hydration and slow strength development.
• Type V is used when concrete must resist high sulfate concentration in soil and
groundwater.
Portland cements are usually gray, but a more expensive white Portland cement
(generally within the Type I or II designations) can be obtained by processing only raw
materials with veiy low iron and transition-elements content.
Blended hydraulic cements are produced by intimately blending two or more types of
cementitious material. Primary blending materials are Portland cement, ground
granulated blast-furnace slag, fly ash, natural pozzolans, and silica fume. These
cements are commonly used in the same manner as Portland cements. Blended
hydraulic cements conform to the requirements of ASTM C-595. ASTM C-595 cements
are as follows: Type IS-Portland blast-furnace slag cement, Type IP and Type P-Portland-
pozzolan cement, Type S-slag cement, Type I (PM)-pozzolan modified Portland cement,
and Type I (SM)-slag modified Portland cement. The blast-furnace slag content of Type
IS is between 25 percent and 70 percent by mass. The pozzolan content ofTypes IP and
P is between 15 percent and 40 percent by mass of the blended cement. Type I (PM)
contains less than 15 percent pozzolan. Type S contains at least 70 percent slag by
mass. Type I (SM) contains less than 25 percent slag by mass. These blended cements
may also be designated as air-entraining, moderate sulfate resistant, or with moderate
or low heat of hydration. The most common blended cements available are Types IP and
IS. The United States uses a relatively small amount of blended cement compared to
countries in Europe or Asia. However, this may change with consumer demands for
products with specific properties, along with environmental and energy concerns.
Expansive cements are hydraulic cements that expand slightly during the early
hardening period after setting. They meet the requirements of ASTM C-845 in which it
is designated as Type E -1. Although three varieties of expansive cement are designated
in the standard as K, M, and S, only K is available in the United States. Type E-l (K)
contains Portland cement, anhydrous tetracalcium trialuminosulfate, calcium sulfate,
and uncombined calcium oxide (lime). Expansive cement is used to make shrinkage-
compensating concrete that is used (1) to compensate for volume decrease due to drying
1-2
-------
shrinkage, (2) to induce tensile stress in reinforcement, and (3) to stabilize long-term
dimensions of post-tensioned concrete structures. One of the major advantages of using
expansive cement is in the control and reduction of diying-shrinkage cracks. In recent
years, shrinkage-compensating concrete has been of particular interest in bridge deck
construction, where crack development must be minimized.
Natural cement is an hydraulic cement produced by calcining argillaceous limestone
below sintering temperatures. Natural cement may be specified by ASTM C-10. It was
used primarily during the 19th century and early 20th century, but largely disappeared
after about 1910, as Portland cement became more popular and began dominating the
market.
Aluminous cement or calcium aluminate cements are hydraulic cements made from
limestone and Bauxite. These cements are principally used in refractory applications.
These cements are produced in tiny quantities with just a few manufacturers worldwide
(USGS, 2005).
Cements can also be specified by performance per ASTM C-1157 and include the
following: Type GU hydraulic cement for general construction, Type HE-high-early-
strength cement, Type MS-moderate sulfate resistant cement, Type HS-high sulfate
resistant cement, Type MH-moderate heat of hydration cement, and Type LH-low heat
of hydration cement. These cements can also be designated for low reactivity (option R)
with alkali-reactive aggregates. Performance based standards are not prescriptive with
respect to composition as in ASTM C-1157 or ASTM C-595, but are inclusive of cements
falling within these standards.
The common industry practice, and that of the U.S. Geological Survey (USGS), includes,
within the Portland cement designation, a number of other cements not within ASTM
C-150, which are composed largely of Portland cement and are used for similar
applications (e.g., concrete) (USGS, 2005). These include blended cement, block cement,
expansive cement, oil well cement, regulated fast setting cement, and waterproof
cement. Plastic cements and Portland-lime cements are grouped within masonry
cement, hydraulic cements for use in mortars for masonry construction. Because
Portland cement accounts for approximately 95 percent of the cement industry's total
production (van Oss, 2008), and because the costs and trends of this industry sector
can be adequately captured by describing the market processes associated with the
production, distribution, and use of Portland cement, in this work the focus is on
Portland cement. In 2006, Portland cement's market share in the U.S. was 94 percent,
while masonry cement's market share comprised the remaining 6 percent (USGS,
2007a).
1.1.2 Overview off the Cement Manufacturing Process
Portland cement is produced from raw materials such as limestone, chalk, shale, clay,
and sand. These raw materials are quarried, crushed, finely ground, and blended to the
correct chemical composition. Small quantities of iron ore, alumina, and other minerals
may be added to adjust the raw material composition. The fine raw material is fed into
a large rotary kiln (cylindrical furnace) where it is heated to extremely high temperatures
(about 2640 °F [about 1450 °C]). The high temperature causes the raw material to react
and form a hard nodular material called "clinker". Clinker is cooled and ground with
1-3
-------
approximately 5-percent gypsum and other minor additives to produce Portland cement.
The main steps in the cement manufacturing process are illustrated in Figures 1-1 and
1-2, which show the wet process and the diy process with cyclone preheater,
respectively. The schematic for a precalciner kiln would be veiy similar to that shown
in Figure 1-2, with the addition of a calciner vessel.
The heart of the clinker production process is the kiln, which can be rotary or vertical
shaft designs. Rotary kilns are commonly used in the U.S. and elsewhere. These kilns
are 6-8 m in diameter and 60 m to well over 100 m long. The kilns are set at a slight
incline and rotate at 1 to 3 revolutions per minute. The kiln is fired at the lower end and
the feed materials move toward the flame as the kiln rotates. The materials reach
temperatures between 1400-1500 °C in the kiln. Three steps occur with the raw material
mixture during pyroprocessing. First, all moisture is driven off from the materials. Then
the calcium carbonate in limestone decomposes into carbon dioxide (CO2) and calcium
oxide (free lime) during calcination. Finally, the lime and other minerals in the raw
materials react to form calcium silicates and calcium aluminates, the main components
of clinker.
1-4
-------
Feed Bins
'wOrrective fcfetenals
Raw Materials
Crushing Plant (s)
. Electrostatic
ledustin
Preclpiator
~assifier
Slurry Mi
Generator
Slurry Bassin
Mneral
Clinker Cooler Gypsum Components
Dispatch Packing Machine Bag Palletisation
C assifier
Lenient
Figure 1-1. Schematic of the Wet Cement Process
Source: CEMBUREAU, 1999
1-5
-------
Fo«d 0m
3ainp»ro Siaton
*1
Comctiva Mjlculr
liau Nbtcnalj llorajc ar4 Pre5jk Q>:pr.ch Itonj Midinc
Oyp?UT Comporero
Lint
'.UiiPK'
Urcm 1,111
Figure 1-2. Schematic of the Dry Cement Process with Cyclone Preheater
Source: CEMBUREAU. 1999
1-6
-------
1.1.3 Kiln Types and Their Use
Rotary kilns are broadly categorized as diy- and wet-process kilns, depending on how
the raw materials are prepared. Wet-process kilns are fed raw material sluny with
moisture content ranging between 30 and 40 percent. A wet-process kiln needs
additional length to evaporate the water contained in the raw material feed. Nearly 33
percent additional kiln energy is consumed in evaporating the water in the sluny.
In dry-process kilns, raw material is fed as diy powder. There are three major variations
of dry-process kilns in operation in the U.S.: long diy kilns, preheater kilns, and
preheater/precalciner kilns. In preheater kilns and preheater/precalciner kilns, the
early stages of pyroprocessing occur before the materials enter the rotaiy kiln. Preheater
and preheater/precalciner kilns have higher production capacities and greater fuel
efficiency compared to other types of cement kilns. Table 1-1 shows heat input in terms
of millions of British Thermal Units (MMBtuJ/ton1 of clinker for various types of kilns.
As the data clearly demonstrate, preheater/precalciner kilns provide greater fuel
efficiency. The replacement of wet and (certain) dry process kiln capacity with modern
kiln processes can yield, theoretically, substantial reductions in fuel use due to fuel
efficiency gains. As the industry moves toward more efficient processes, replacement of
wet and long diy process capacity with more efficient kiln process technologies is
expected.
Table 1-1. Typical Average Heat Input by Cement Kiln Type
Kiln Type
Heat Input,
MMBtu/ton of clinker
Wet
6.0
Long Diy
4.5
Preheater
3.8
Preheater/Precalciner
3.3
Source: EPA, 2007 (Table 3-3)
As expected, a recent trend in the cement sector has shown the replacement of lower
capacity, inefficient wet and long dry kilns with bigger and more efficient kilns. This
trend is expected to continue. In the U.S., the overall number of kilns decreased by 11
percent from 1995 to 2004. During the same period, total clinker production capacity
increased by 18.6 percent. Portland Cement Association (PCA) data show that average
kiln capacity also increased by 27 percent, from 405,000 to 556,000 tons per year
between 1995 and 2004. The number of kilns operating in the U.S. in 2005 compared
with the number of kilns in operation in 2009 is shown in Table 1-2. The trend in kiln
design and average kiln capacity is shown in Figure 1-3.
1 PLEASE NOTE: for the purposes of this document, short tons will be referred to simply as "tons" and represents 2000 lbs. A
unit conversion table is provided in the front matter of this document.
1-7
-------
Table 1-2. Number of Kilns by Kiln Type in the U.S. in 2005 and 2009
Kiln Type
Number of Kilns (2005)
Number of Kilns (2009)
Wet
50
42
Dry
48
32
Preheater
36
29
Precalciner
47
58
Source: PCA, 2006 and PCA, 2009a
225
200 ¦
175 ¦
150 ¦
125 ¦
(/>
= 100 H
5 75 H
X!
i 50 ¦
25 ¦
1995 1996 1997
1999 2000 2001 2002 2003 2004 2005 2006 2007
I I Dry Kilns I iWet Kilns * Average Kiln Capacity
600
¦¦ 500
¦¦ 400
¦¦ 300
¦¦ 200
en
TJ
c
3 re
o ^
£ *3)
o
re
Q_
re
O
o
O)
re
-------
in 2005 are shown in Figure 1-4. The cement manufacturing sector in the U.S. is
concentrated among a relatively small number of companies, many of which are owned
by or are subsidiaries of foreign companies (USGS, 2007b). Together, ten companies
accounted for about 80 percent of the total cement production in U.S. for 2005 (USGS,
2007b). California, Texas, Pennsylvania, Florida, and Alabama are the five leading
cement-producing states, which accounted for about 43 percent of the total production
in 2005 (USGS, 2007b: Table 3).
In 2009, PCA projected a capacity expansion of 27 million tons between 2008 and 2013,
an 18-percent increase in existing capacity from 2006. PCA's projected capacity
expansions are to come from 23 kilns; five came on-line in 2008 and 18 more are
expected to come on-line between 2009 and 2013 (PCA, 2009b). The investment in these
projected capacity expansions is projected at $6.9 billion. Note that a typical project for
a new facility (greenfield) from ground breaking to startup has a timeframe of about 2
to 3 years. Building a new kiln in an existing facility (brownfield) can take approximately
1 year (EPA, 2007). If permitting for mining and (re)construction is accounted, a typical
project can take as long as 4 to 6 j^ears for a greenfield facility and up to 3 years for a
brownfield facility (Andover Technology Partners, 2009a).
Cement Plant Locations
\
• CernentPI ant Locations >, ^ f
Largest Cement Import Custom Districts \
A Cementlrn port Custom Districts
Mississippi-Hudson River
Source: EFA20D2-2xe Dal,
I
%
Figure 1-4. Portland Cement Plant Locations
2.2.5 Imports of Portland Cement in the U.S.
Portland cement is not only produced and consumed domestical^, but it is also traded
internationally. In 2005, the U.S. (not including Puerto Rico) produced 94 million metric
tons of cement (USGS, 2007b: Table 3) and imported 34 million metric tons of hydraulic
1-9
-------
cement and clinker (USGS, 2007b: Tables 3 and 18). The level of imports to the U.S. is
highly cyclical, with domestic producers importing primarily when domestic plants are
at full capacity and cannot meet excess demand. Generally, imported cement and
clinker make up 20 to 27 percent of domestic cement consumption. In 2005, total
imports of cement and clinker (especially clinker) increased, owing to continued high
demand; imported cement accounted for about 24 percent of the total cement sales in
the U.S. (USGS, 2007b).
In 2005, the ten leading international cement and clinker suppliers to the U.S. were, in
descending order, Canada, China, Thailand, Greece, the Republic of Korea, Venezuela,
Mexico, Colombia, Taiwan, and Sweden. The ten busiest ports of entry within customs
districts existing in 2005 were, in descending order, New Orleans, Tampa, Los Angeles,
Houston-Galveston, San Francisco, Miami, Seattle, Detroit, New York, and Charleston
(South Carolina) (USGS, 2007b). Table 1-3 shows the major customs districts for
hydraulic cement and clinker imports in the U.S. Figure 1-5 shows the imports of clinker
and cement from 1998 to 2008.
Table 1-3. Largest Hydraulic Cement and Clinker Import Custom Districts in the U.S. in
2005
Import of Hydraulic Cement and Clinker Percentage
Custom District
thousands of
thousands of
of Total
U.S.
Imports
metric tons
tons
New Orleans, LA
4,095
4,514
12.31
Tampa, FL
3,478
3,834
10.46
Los Angeles, CA
3,053
3,365
9.18
Houston-Galveston, TX
2,619
2,887
7.87
San Francisco, CA
2,363
2,605
7.10
Miami, FL
2,265
2,497
6.81
Seattle, WA
1,489
1,641
4.48
Detroit, MI
1,317
1,452
3.96
New York, NY
1,264
1,394
3.80
Charleston, SC
1,102
1,215
3.31
Total
23,046
25,404*
69.29%
* rounding, original data in metric tons
(Source: USGS, 2007b (Table 18)
1-10
-------
40000
35000 -
30000 -
25000 -
« 20000
c
as
S 15000 4
10000 -
5000 -
~ Cement ~ Clinker
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Figure 1-5. Imparts of Clinker and Cement from 1998 to 2008 (PCA, 2009a)
1.1.6 Cement Demand Centers
Because of the relatively high transportation costs, the U.S. cement industry is
structured around state-specific cement demand centers. PCA reports that the vast
majority of cement produced in the U.S. is being transported less than 300 miles by
truck due to cement's low value by weight and high cost of transport (PCA, 2005).
However, cement may be transported over longer distances, especially when the less
expensive rail and water transportation modes are available (APCA, 1997).
1.2 Emissions from the U.S. Cement Industry and Applicable
Regulations
Criteria pollutants, hazardous air pollutants (HAPs), and CO2 are released during
cement manufacturing. Nitrogen oxides (NOx) emissions from cement kilns result
primarily from the combustion process: oxidation of fuel nitrogen (fuel NOx) and the
oxidation of nitrogen in the combustion air (thermal NOx). EPA's 2005 National Air
Toxics Assessment (NATA) Inventory reports that cement kilns released 181,000 metric
tons (200,000 tons) of NOx emissions from the combustion of fuels (EPA, 2009).
Sulfur dioxide (SO2) emissions2 from cement kilns result from the combustion of sulfur-
bearing compounds in coal, oil, and petroleum coke, as well as sulfur compounds in
raw materials. Sulfur in the fuel will oxidize to SO2 during pyroprocessing and a
significant amount is likely to be captured in the form of sulfates as the flue gas passes
through the calcination zone. Compared to long dry and wet kilns, preheater and
precalciner kilns tend to be more effective at capturing fuel-gene rated SO2. Accordingly,
2 Small amounts of sulfur trioxide (SO3) may be released in addition to bulk SO2 but the SO3 emissions are treated as SO2 for
computational purposes.
1-11
-------
oxidation of sulfur in the feed materials is likely to be the major component of total SO2
emissions. The 2005 NATA Inventory reflects that cement kilns released 133,000 metric
tons (147,000 tons) of S02 emissions in 2005.
Particulate matter (PM) emissions result from quarrying operations (the crushing and
grinding of raw materials and clinker) as well as kiln line operations. The 2005 NATA
Inventory estimates show that, in 2005, cement kilns released 10,000 tons of PM10
emissions. The cement industry is also a source of HAPs (e.g., hydrochloric acid vapor
and chlorine), as well as metals including but not limited to mercury, antimony,
cadmium, and lead (EPA, 2008).
The cement manufacturing process is also a source of CO2 emissions. CO2 emissions
are a product of the combustion of fuel as well as the calcination of the limestone in the
raw meal. The CO2 from fuel combustion can be calculated from heat input and fuel
characteristics combined with clinker production. The CO2 from calcination can be
calculated taking into account the amount of limestone used as a source of clinker
calcium. Limestone currently is the predominant source of calcium for the clinker. Pure
limestone would produce 0.44 tons of CO2 for every ton of limestone completely calcined
to calcium oxide. Substitute materials may be used in lieu of limestone with the effect
of reducing the CO2 emissions from clinker production. The CO2 emissions reduction
would be proportional to the amount of substitute materials added. Detailed discussion
of CO2 emissions from cement kilns can be found in Appendix A.
Multiple regulatory requirements to reduce criteria air pollutants and HAPs emissions
currently apply to the cement industry sector. The New Source Performance Standards
(NSPS) and the National Emissions Standards for Hazardous Air Pollutants (NESHAPs)
are two of the federal requirements that apply to cement facilities. Additionally, state
and local regulatory requirements might apply to individual cement facilities depending
on their locations. In 2008, 44 cement facilities were located within ozone (O3)
nonattainment areas (NAA), while 20 facilities were within PM2.5 nonattainment areas.
Seventeen facilities were found to be located in (or within 30 miles of) Class 1 areas, as
shown in Figure 1-6. Class 1 areas are areas of special natural, scenic, recreational, or
historic (national or regional) value for which the Prevention of Significant Deterioration
(PSD) regulations provide special protection.
1.3 Overview of U-ISIS
U-ISIS, a sector-based dynamic programming model, will facilitate the analyses of
emission reduction strategies for multiple pollutants while taking into account plant-
level economic and technical factors such as the type of emission units (for cement -
kiln), associated capacity, location, cost of production, applicable controls, and their
costs. For each of the emission reduction strategies under consideration, the model is
able to provide information on the following:
• Optimal (least cost) industry operation
• Cost-effective controls to meet the demand for cement
• Emission reduction requirements over the time period of interest
1-12
-------
U-ISIS incorporates multiple industries within a multi-market, multi-product,
multipollutant, and multi-region emissions trading framework. The objective function
in U-ISIS maximizes total surplus and uses an elastic formulation of the demand
function to estimate area under the demand curve,
Cement Facility Locations
• Cement Pla nt Locatio ns
Largest Cement Import Custom Districts
~ Cement I mport Custo m Distri cts
- Mississippi-Hudson River
| Class I Areas
Ozone NonAttainment Areas
PM2.5 NonAttainment Areas
Source: EPA2002-2006 Data
1
Figure 1-6. Portland Cement Facilities and O3 NAA, PM2.5 NAA, and Class 1 Areas
The U-ISIS code is written in the General Algebraic Modeling System (GAMS) language,
Input data, organized in various spreadsheets of a Microsoft Excel workbook, are passed
on to the GAMS files. These input data consist of an industry database, which provides
unit-level production, capacit}^, production cost, and emissions information. The
controls database provides information regarding applicable air pollution control
technologies and their cost and emission control characteristics. A policy module is used
to specify various parameters of interest to the policy analyst, such as emissions cap,
emission reduction scenarios, and discount rate. The input data, control data, and
policy parameters are then transmitted to the optimization part of the U-ISIS model,
where they are used to solve the selected base and policy cases. After solving, the results
are post-processed to calculate values of various outputs of interest. The output data
are exported to Excel spreadsheets for further analyses and graphical representation of
selected results.
Within an industrial sector, generally emissions arise from four pathways: (1) on-site
emissions due to combustion of fossil fuels for energy at plants, (2) on-site emissions
due to processing of certain raw materials (e.g., limestone calcination in cement plants,
non-energy uses of fossil fuels in chemical processing and metal smelting), (3) off-site
1-13
-------
emissions due to combustion of fossil fuels at power plants to generate the electricity
needed by the industrial sector, and (4) overseas emissions associated with imports.
These pathways are depicted in Figure 1-7.
Pollution Reduction Policy
(Rate-based, Cap-and-trade,
Emissions Taxes)
Water
Fuel
(Fuel Substitution/
Reduction)
Electricity
(Change in Fuel, Power
Plant Emissions)
Raw Materials
(Raw Material Substitution)
Industrial Sector
(Cement, Pulp & Paper, Power and
Iron & Steel etc.)
Improvements
Products
(Production Substitution)
J
Market(s)
Pollution Controls
(NOx, C02, S02, VOC, PM, ...)
Air Emissions
Waste
Imports
(Oversees Emissions)
Figure 1-7. An Integrated View of Pollution Generation Pathways, Emissions Abatement
Approaches, and Multimedia Impacts for an Industrial Sector
Also shown in Figure 1-7 are the potential options for abating emissions from industrial
sectors and multimedia impacts. The options shown in green are pollution prevention
measures and the ones in red are mitigation measures. Clearly, the integrated picture
presented in Figure 1-7 makes a compelling case for considering commodity
production/supply activities along with emissions while developing holistic emission
reduction strategies. While developing the U-ISIS framework, care has been taken to
build the emission pathways and abatement options shown in Figure 1-7. Example
emission reduction policies that can be evaluated using U-ISIS are:
• Criteria pollutants (NOx, SO2, PM, carbon monoxide (CO)) -emission limits and/or
cap-and-trade
• HAPs (e.g., mercury (Hg), hydrochloric acid (HC1)) - emission limits
• CO2 - cap-and-trade and/or emission taxes
• Long and short time horizons: CO2 (decades), criteria pollutants (annual)
• Regional or national requirements
1-14
-------
1.4 References for Chapter 1
Andover Technologies Partners (2009a). Memorandum: NOx, SO2 and CO2 Emissions
from Cement Kilns (Emissions Memo), from Jim Staudt to Ravi Srivastava,
Samudra Vijay, Elineth Torres. Dated March 10, 2009 (see Appendix A of this
document).
APCA (1997). Comments on EPA's Draft Economic Analysis of Air Pollution Regulations
for the Portland Cement Industry (May 1996), prepared for American Portland
Cement Alliance by Environomics, January 29, 1997.
ARCADIS (2010). Development of U.S. EPA Industrial Sector Integrated Solutions (ISIS)
Model, Quality Assurance Project Plan. Prepared under contract number EP-
C-09-027 for the U.S. Environmental Protection Agency, Air Pollution
Prevention and Control Divisions, Research Triangle Park, NC. February 3,
2010.
CEMBUREAU (1999). Best Available Techniques for the Cement Industry, CEMBUREAU
Report, The European Cement Association, December 1999.
D/1999/5457/December, Brussels, http:/ Zwww.cembureau.be
EPA (2007). Alternative Control Techniques Document Update - NOx Emissions from
New Cement Kilns. EPA-453/R-07-006, November 2007.
http://www.epa.gov/ttn/catc/dirl/cement updt 1107.pdf, accessed
October 21, 2008.
EPA (2008). AP 42, Fifth Edition, Compilation of Air Pollutant Emission Factors, Volume
I, Chapter 11: Mineral Products Industry.
http: / /www.epa. gov/ttn/chief/ap42/chl 1/fraal/cl ls06.pdf, accessed
October 21, 2008. Complete document:
http: / /www.epa. gov/ ttn/chief/ ap42 /. accessed October 21, 2008.
EPA (2009). EPA 2005 National Air Toxics Assessment Inventory. U.S. Environmental
Protection Agency. October 2009.
NAS (2004). Air Quality Management in the United States. National Research Council
(U.S.), Committee on Air Quality Management in the United States, National
Academies Press, Washington, 2004.
http://books.nap.edu/catalog.php9record id= 10728. accessed October 21,
2008.
PCA (2005). Letter from David S. Hubbard, Director, Legislative Affairs, Portland Cement
Association. RE: Hours of Service of Drivers; Proposed Rule (Docket Number
FMCSA-2004-19608), March 10, 2005.
http://www.cement.org/exec/DHQS%20Comments percent203.10.05.pdf.
accessed October 21, 2008.
PCA (2006). U.S. and Canadian Portland Cement Industry: Plant Information Summary.
Portland Cement Association, Skokie, IL, 2006.
PCA (2009a). PCA Annual Yearbook 2009. North American Cement Industry. Portland
Cement Association.
PCA (2009b). PCA Capacity Report Flash Report. Portland Cement Association's
Economic Research Department. Updated October 19, 2009.
1-15
-------
USGS (2005). Background Facts and Issues Concerning Cement and Cement Data. U.S.
Geological Survey, Open-File Report 2005-1152.
http://pubs.usgs.gOv/of/2005/l 152/2005- 1152.pdf, accessed October 21,
2008.
USGS (2007a). Mineral Commodity Summaries, pp. 40-41, January 2007,
http://minerals.usgs.gov/minerals/pubs/commodity/cement/cemenmcs07
¦ pdf, accessed September 24, 2009.
USGS (2007b). 2005 Minerals Yearbook: Cement. U.S. Geological Survey, p. 16.2,
February 2007,
http://minerals.usgs.gov/minerals/pubs/commodity/cement/cemenmyb05
.pdf, accessed September 24, 2009.
van Oss, H.G. (2008). Personal communication from Hendrik G. van Oss, USGS, to
Elineth Torres, U.S. EPA, on July 7, 2008.
1-16
-------
Chapter 2
U-ISIS Framework
U-ISIS is a sector-based linear programming model that can help analyze optimal sector
operations for meeting demand and pollution reduction requirements over specified time
periods. The objective in U-ISIS simulation is to maximize total surplus (see Figure 2-1)
over a time horizon of interest for an industry. The total surplus concept has long been
a mainstay of social welfare economics because it takes into account the interests of
both consumers and of producers (Samuelson and Nordhaus, 1977).
Price
Inverse
Supply Curve
Consumer
surplus
Equilibrium
Producer
surplus
InverseDemand
Curve
Quantity
Qi Qe
Figure 2-1. Total Surplus In a Market
In a market at competitive equilibrium, without exogenous factors, the total surplus can
be thought of as composed of producer surplus and consumer surplus. Using Figure 2-
1, the producer surplus corresponding to a quantity Q of a commodity is the difference
between the gross revenue and the inverse supply curve. Gross revenue is simply the
product of the price and the quantity consumed. Similarly, the consumer surplus
corresponding to a quantity Q is given by the area under the inverse demand curve up
to that quantity minus the gross revenue. The consumer surplus is the cumulative
opportunity gain of all consumers who purchase the commodity at a price lower than
the price they would have been willing to pay. It is evident from Figure 2-1 that the total
surplus is maximized exactly when Q is equal to the equilibrium quantity Qe. This result
allows determination of the demand quantity at equilibrium, where total surplus is
maximized, as the cost of production meets the inverse demand curve.
2-1
-------
When the quantity consumed is less than the optimum QE(e.g., Qi) due to some market
disturbance such as a policy or regulation change, the consumer pays a higher price Ci
resulting in a reduction in consumer surplus. The inverse supply curve shifts as a result
of this disturbance. At a new equilibrium the marginal total cost will increase from Pi of
the base case to Ci at the lower quantity Qi. The total surplus shrinks resulting in a
welfare loss, represented by the lighter shaded area in Figure 2-1. The producer surplus
changes from the base case as a proportion of total surplus, but some of this surplus
may be diverted from the producer by the form of the policy or regulation change. The
framework of the U-ISIS model does not proportion the total surplus into consumer and
producer surplus, but calculates the total surplus from the total benefit, i.e., the total
area under the inverse demand curve less the area under the inverse supply curve for
the quantity of interest.
The general concept of spatial price equilibrium (SPE) in a network, where the mutual
influences of production, transportation, and consumption patterns are given full
consideration, has been developed over the past 6 decades. In SPE network models,
interregional economies are simulated by finding the balance of demand, supply, and
trade that will result in competitive market equilibrium among the regions. Enke (1951)
first demonstrated how the cost of transportation might be included in an equilibrium
analysis of spatially separated markets by means of analogy with resistance to the flow
of current in an electric circuit. Shortly after Enke, Samuelson (1952) analyzed
interregional flows of commodities and market equilibrium using a linear programming
formulation. In this type of formulation, the equilibrium for each market of a sector is
equivalent to the quantities and prices that result while maximizing the sum of
consumer and producer surpluses for each market of the sector. This sum is referred to
as the total surplus or net social payoff of the sector; McCarl and Spreen (1980) provide
interpretation and justification. The linear programming formulation of the SPE problem
was developed by Duloy and Norton (1975).
PAPER
Figure 2-2. Modular Architecture of U-ISIS
2-2
-------
The Industry inputs part of U-ISIS allows users to enter industrial sector data including
the number of production facilities, distance from production facilities to demand
center, production capacity, associated costs (material costs, operations and
maintenance costs, etc.), fuel types and costs, emissions sources and intensities, etc.
The Market inputs part of the model includes historical and projected nationwide
commodity consumption, commodity imports and exports. The Others inputs allows
users to define discounts rates, electricity parameters, escalation rates, economic life,
and technologies and emissions abetments factors. The Policy inputs allows users to
define policy related parameters such as the amount of control required and the total
quantity of emissions allowed. The user can specify the policy horizon (time period) to
be used for the model runs. Policies may be simulated over long and short time horizons,
such as a CO2 policy that occurs over a decadal time-frame, and a criteria pollutant
policy that occurs on an annual-time frame. The U-ISIS model is also capable of
evaluating requirements at a regional or national-scale.
2.1 Objective Function
The objective function of the model is to maximize the net present value total surplus
(by the equivalent function, minimizing the negative of the total surplus) for the given
sector of interest over a selected time horizon. The total costs, as calculated by the
algorithm, approximate the area under the inverse supply curve. Components of total
costs include production cost, transportation cost, import cost, control cost, energy
efficiency cost, and emission charge. The total benefit - which includes total costs,
producer surplus, and consumer surplus - is calculated from the total area under the
inverse demand curve. Each element is corrected to net present value by applying a
discount factor for each year within the time horizon based on a user supplied discount
rate. The negative of the total surplus is calculated by subtracting the total benefit from
the total costs.
Elements of total costs include:
1) Production cost - obtained for each cement production unit. Each unit's
production cost takes into account the factor input costs of raw material, labor,
energy, and other cost components.
2) Transportation cost - cost of transporting from supply center to the demand
center. Production from each supply center may be transported to any demand
center. Distance from each supply center to each demand center is incorporated
in the industry inputs.
3) Import cost - calculated by multiplying the quantity of imported goods by the
import price for each country of origin and adding any handling and other
associated costs. All imports arrive at the import terminals and incur
transportation costs to reach each demand center; distances from import
terminals to each demand center are incorporated in the industry inputs module.
4) Control and energy efficiency costs - includes the capital and variable cost of
installing controls and energy efficiency options, to achieve any emission
reduction targets governed by the constraints.
2-3
-------
5) Emission charge - added if any allowance price is given for the pollutants.
The objective function is:
Minimize z = ^discount factor(time) • production cost(time, production unit, final product) +
t,i,n
y discount factor(time)* transportation cost(time, demand center, final product) +
t,dc,n
y discount factor(time)*import cost(time, import district, final product) +
t,id,n
^discount factor(time) • control cost(time, production unit, control option) +
t,i,k
^discount factor(time) • energy efficiency cost(time, production unit, efficiency option) +
t,i,ee
^discount factor(time)* allowance price(time,pollutant) • total emissions(time, pollutant)-
tj
y [discount factor(time)* total benefit(time, demand center, final product)]
!,dc,n
Where the quantities appearing in the equation above are defined as:
• discount factor(time) is the factor to correct costs from year t to net present
value,
• production cost(time, production unit, final product) is the production cost in
year t for production unit i of final product n,
• transportation cost(time, demand center, final product) is the cost for year t
of transporting the product n to demand center dc,
• import cost(time, import district, final product) is the cost for year t of
importing foreign product n in import district id,
• control cost(time, production unit, control option) is cost for year t to control
production unit i using control technology k,
• energy efficiency cost(time, production unit, efficiency option) is the cost for
year t for production unit i to use the energy efficiency measures ee,
• allowance price(time, pollutant) is the user defined allowance price for
pollutant j in year t
• total emissions(time, pollutant)is the total emissions of pollutant j duringyear
t from the sector, and
• total benefit(time, demand center, final product) is the area under the
demand-price curves for product n for all markets dc.
Total costs, as calculated by the algorithm, approximates the inverse supply curve by
filling demand from the lowest cost product incorporating production or import costs,
transportation, and policy measures through consecutively higher cost product until
demand is satisfied. Demand is satisfied when the demand price no longer exceeds the
supply cost. The user chooses a range of interest centered on the expected demand for
2-4
-------
demand center and production year; model default is 0.5 to 1.5 times the expected
demand. Demand in this range is divided into a user defined number of steps or
intervals; the model default is 100 st€ps. The inverse demand curve is used to determine
the demand price at the midpoint of each demand step user supplied data (e.g.,
elasticity] in each demand region using a constant elasticity of demand:
W>-w(£)7 i2-1)
Where:
D is the demand for the commodity with corresponding price P{D),
° is the elasticity of demand relative to price, and
DO and F0 are the initially-specified demand quantity and price, respectively.
The total benefit is calculated based on a constant elasticity of demand model in the
same stepwise fashion as illustrated in Figure 2.3. The benefit within the demand range
considered by the user, from D,-,„ to the final demand quantity, is estimated by the
product of price at the midpoint of each step and the width of the step. The benefit
associated with demand from zero to Drain is estimated by the product of Drain and the
demand price of the first step of the range.
Price
Figure 2-3. Stepwise Integration of the Inverse Demand Curve
Production costs are a combination of fixed cost, in the form of capital recovery costs,
and variable costs. Capital recovery usually depreciates the cost of new production
capacity over the economic life of the additional capacity using a user-defined interest
rate for capital expenses. However, the user will have the option to add fixed costs, if
any, for an existing production unit.
2-5
-------
The U-ISIS model includes constraints for ensuring that production capacity changes
occur in a realistic way. Production is modeled for five types of units: existing production
units, expansion units, replacement units, projected units and new production units.
Existing production units are units currently installed and capable of producing
product. Expansion units are the units associated with increasing production capacity
at an existing production unit. Replacement units are production units built to retire
existing production units and replace with new production units. New production and
projected units represent entirely new production capacity.
Variable production costs include raw material costs, operations and maintenance costs
(O&M), labor costs and electricity costs provided by the user on a unit of production
basis. The model adds fuel cost, water cost and solid waste cost. Fuel cost is calculated
as the product of energy intensity of production and fuel cost for each type of fuel used.
Similarly, water cost is the product of the water used in production and the price of
water per unit volume. Solid waste cost is the product of the amount of solid waste
produced and the price of solid waste disposal.
Transportation costs are the costs associated with moving production from factories to
the demand center and the costs of moving imports from the import district to the
demand center. The variable cost of transporting each unit of supply from the factory or
import district to the demand center is supplied by the user. The transportation cost is
the sum of the product of variable transportation costs for each factory or import district
and the supply from that factory or import district.
Imports cost to each import district is the product of imported quantity and the cost of
importing product. The imported quantity to each import district is iteratively
determined from the marginal cost of domestic production, at the high cost production
facility, and the total cost associated with the imports inclusive of transportation. Costs
of imports is the sum of the import price to the import district, insurance and freight to
the import district, and handling costs at each import district. The import price is
determined from a constant elasticity of supply curve for each import district based on
user-supplied information.
Similar to the added production, costs for both controls and energy efficiency measures
consist of both capital recovery costs and variable costs. Both farther modify the energy
intensity of production, and therefore, the fuel cost as well. Both measures are
amortized over the expected lifetime of the modification.
The cost of emissions is determined for each pollutant as the product of the emission
and allowance price for the respective emission. No emissions costs are associated with
imports. These costs are consistent with a user-supplied emission tax.
2.2 Constraints
The objective function is minimized with regard to the constraints described in the
following sections to arrive at the optimal solutions.
Consumption & Supply for each demand center; the total supply has to be greater than
or equal to consumption in the given time period. Supply can be comprised of local
production, import from other regions, and foreign import. U-ISIS provides fall flexibility
2-6
-------
to determine demand centers, imports/exports terminals, commodities quantity and
price, and associated transportation costs. Total domestic consumption for a commodity
can be satisfied by domestic production and foreign imports as follows:
2 Production Quantity + £ Imports Quantity — £ Exports Quantity > £ Consumption (Demand) (2-2)
where import/export quantity is limited to terminal capacity. Production of a commodity
is limited to cement plants' availability. Plant availability can be restricted by resources
availability such as fuel and raw materials availability and capacity. For instance, energy
consumption by a plant can only be picked from the fuels available at the location of
production. While in reality switching of fuels, e.g., from coal to natural gas, can be
complicated requiring capital investment in the infrastructure. For the purpose of this
model, the fuels are assumed to be perfectly substitutable without any additional cost
of switching.
Production [ ~ ^ (2-3)
(< Fuel availability v '
Emissions pathways are included in the U-ISIS framework. The U-ISIS framework
includes algorithms to account for tracking multiple pollutant streams associated with
uncontrolled emissions, controlled emissions, pollution prevention from process
modifications and energy efficiency measures, and any controls-related effects. For a
given pollutant, total emissions have to be limited to emission limits specified by the
exogenous policy constraints on emissions. If the policy being analyzed allows for
banking of emissions, then the banking equation enables banking of allowances for
future use.
Transportation of goods and commodities from a production unit is limited by the lower
of the two - the production capacity of the unit or the transportation capacity from a
production unit to all demand centers, if specified. Certain other constraints on
transportation can limit the transportation network based on empirical studies.
Imports and Exports quantity on each terminal is limited by the terminal capacity, but
U-ISIS provides full flexibility to customize assumptions including changes in quantity,
imports/exports prices and terminal locations.
Emissions Abatement Approaches in U-ISIS are categorized in three abatement
approaches: process modifications and upgrades, raw material and/or fuel substitution,
and mitigation technologies. For each emission abatement approach, where possible,
information is included in the U-ISIS model on capital cost, fixed operating cost, variable
operating cost, emission reduction performance for all of the pollutants, impacts on fuel
and/or raw material use, impact on electricity consumption, byproduct generation and
cost, and impact on water use parameters.
Policy Parameters in the U-ISIS Framework allows the user to select a variety of potential
policy options for evaluation. The user can select from cap-and-trade policy, emissions
charge policy, or rate-based policies. In a cap-and-trade policy scenario, separate caps
on pollutants of interest can be specified. The user has the option to run a cap-and-
trade policy scenario with or without banking of emissions. Further, a cap-and-trade
policy scenario can include de minimus requirements, where the user defines a
2-7
-------
minimum level of emission reduction required for each emission unit. It is also possible
for the user to input an emission charge for pollutants of interest. Furthermore,
traditional policy scenarios (rate-based policies) with unit-specific emission reduction
requirements specified by the user can be modeled in U-ISIS.
2.3 Optimization and Post Processing
In U-ISIS, the input data are pre-processed to arrive at suitable input parameters for
use in the model equations explained in this chapter. Once the data have been pre-
processed, U-ISIS solves for the appropriate levels of production, imports and controls
required to meet the constraints associated with commodity demand and emissions,
while maximizing total surplus. Once the surplus maximization problem has been
solved, the results are post-processed to obtain parameters and level values of the
variables of interest. The key variables of interest are production level of each production
unit to meet regional demand, level of imports in each region, installation of various
controls, emissions, and various costs. Output data are written in appropriate
worksheets in an Excel workbook and further linked to various plots to enable visual
presentation and analyses of the results.
2.4 U-ISIS Interface
The U-ISIS interface is a single PC-based executable tool with a multi-sector (modular)
approach that provides a user-friendly interface for exploring and comparing various
scenarios of meeting product demand and pollution reduction requirements for an
industrial sector of interest over specified time periods. The U-ISIS interface allows the
user to develop, edit, or delete scenarios for an industrial sector of interest. Both the
web application and the downloadable tool are being developed by EPA's Air Pollution
Prevention and Control Division (APPCD) within ORD. A beta version of the web
application is being developed. Data collection (from published data and from well-
developed existing technologies) is underway to produce and populate data for the web
application to help ensure it will become a robust, production-ready application. Figure
2-4 shows the modeling framework for the cement industry and the types and
interaction between the various parameters involved.
2-8
-------
Onsite
emissions
ROW
Limestone
Clinker Production
Raw Materials
(Crutfilng, CrlnAtg mtl
¦lending)
Fuels Consumption
(Coal, M«,ol,nr*« ate.)
Imports Terminals
(Handhg, Taxes fk
Tracing Costs)
Rotary Klin
Clinker Cooler
Products
Transport
Cement
Production
Domestic
Consumption
Electricity
emissions
Gypsum
NG = Natural Gas
ROW = Rest of the World
Figure 2-4. Cement Modeling Framework
The ultimate goal of the web application is to provide a database for users, both internal
and external of EPA, to explore emissions and least-cost scenarios for supply, suppfy
costs, and emission control costs under various policy options for the industrial sectors
covered in the modules. This web application will link to the database, developed and
maintained by ORD/APPCD. This database, once available to EPA and the public will
need updating on a regular basis to include new and updated information as they
become available. The user will have the option of querying the database, through the
web application, and populating data and information based on their scenario of
interest. With this, the user will have the capability to export the data from the web
application to Excel or in PDF format.
2.4.1 Interface Features
The features of the user interface will include pull-down menus, mouse support and
"point-and-click" activation of many of the features and will be tested using "internal
best test?' procedure. The U-ISIS database will be full}'' protected, such that each user
scenario option will be evaluated individually. Step-by-step instructions for using each
module of the U-ISIS model will be provided to all external users and will be updated as
needed. An}^ user will be expected to have read the U-ISIS Quality Assurance Project
Plan (QAPP) and industry-specific module appendix as well as be familiar with the
model, how it works, and the types of outputs to be generated.
2-9
-------
2.4.2 Interface Data Structure
As shown in Figure 2-2, U-ISIS has a modular architecture and emission, fuel, and
policy parameters relate to each industry-specific module. Input data are organized in
various spreadsheets of a Microsoft Excel workbook to run as a standalone (single
module without interface) version of the U-ISIS model. U-ISIS data are organized in a
Microsoft sql database. U-ISIS code is written in the General Algebraic Modeling System
(GAMS) language. The U-ISIS interface is being programmed in C# and web development
programming software in a graphical and web-based layout. The interface plays an
intermediate role between the U-ISIS model (developed in GAMS) and as an interface to
exchange data. Furthermore, the interface retrieves data from the U-ISIS model and
allows the user to generate tables and graphs of interest.
All data are organized in a Microsoft Access database and in Excel. The U-ISIS interface
communicates with the Microsoft Access database, generates input data sheets, and
transmits to the U-ISIS model for the optimization. Selected data are then pre-processed
in the U-ISIS model to arrive at suitable input parameters for use in the model
equations. After pre-processing the data, U-ISIS solves for the appropriate levels of
production, imports, and controls required meeting the constraints associated with
commodity demand and emissions, while maximizing total surplus. Once the surplus
maximization problem has been solved, the results are post-processed to obtain
parameters and level values of the variables of interest. After solving, the output results
are transferred into a Microsoft Access database. The U-ISIS interface then helps the
user to interpret these outputs in the desired format (tables, graphs, etc.). The
systematic diagram of the interface and U-ISIS engine is shown in Figure 2-5.
INPUTS INTERFACE
— "" *"
-- -
...
General
Inpu
inputs
o
ISIS Database
<=>
» - ^ t ^ £ *
United Statos Environmental Protection Agency
Industrial Sector
Integrated Solutions (ISIS) Model
ISIS Engine
Figure 2-5. Interface of the U-ISIS
2-10
-------
The general input interface of U-ISIS helps the user to develop the required modeling
framework of the industrial sector of interest which includes time horizon (simulation
period) to be used for the model runs, reference year, discount rate, time blocks,
commodity characteristics , emissions types, fuels types, plants types, etc. The general
input interface allows the users to enter three sets (market, facility based and other
data) of Business-As-Usual (BAU) industrial sector data which include historical and
projected nationwide commodity consumption, commodity imports and exports, the
number of production facilities, distance from production facilities to demand center,
production capacity, associated costs (material costs, operations and maintenance
costs, etc.), fuel types and costs, emissions sources and intensities, etc., that the Market
(Inputs) module includes.
The Policy (Inputs) module allows users to define policy related parameters such as
the amount of control required and the total quantity of emissions allowed. The user
can specify the emission reduction percentage of interest, allowance, banking or non-
banking, taxes, minimum reduction levels and policy horizon (time period) to be used
for the model runs. Policies may be simulated over long- and short-time horizons,
such as a CO2 policy that occurs over a decadal time-frame, and a criteria pollutant
policy that occurs on an annual time-frame. The U-ISIS model is also capable of
evaluating requirements on a regional or national-scale.
The goal of the PC-based executable application is to provide data entry and modeling
flexibilities and easy access to U-ISIS modules (multi-sectors) for its users, both internal
and external to EPA, to explore emissions and least-cost scenarios for supply, supply
costs, and emission control costs under various policy options for the industrial sectors
covered in the U-ISIS modules. The user will have the capability to export the outputs
from the interface to Excel, or to text, graphic or PDF format.
The functionality of the interface will ensure that users will be allowed to use
individually chosen general inputs as well as policy inputs and will be able to access the
EPA-hosted database and U-ISIS engine to produce output for the desired type of
analysis for the industrial sectors of interest. The U-ISIS database will be fully secured
and protected, so that each user's scenario option will be evaluated individually.
2-11
-------
2.5 References for Chapter 2
Duloy, J. H.; Norton, R.D. (1975). Prices and incomes in linear programming models.
American Journal of Agricultural Economics. 57(4): 591.600.
Enke, S. (1951). Equilibrium among spatially separated markets: solution by electric
analogue. Econometrica. 19: 40.47.
McCarl, B.A.; Spreen, T.H. (1980). Price endogenous mathematical programming as a
tool for sector analysis. American Journal of Agricultural Economics. 62:
87.102.
Samuelson, P.A. (1952). Spatial price equilibrium and linear programming. American
Economic Review. 42(3): 283.303.
Samuelson, P.A.; and W. Nordhaus. (1977). Economics (17th edition), John Wiley.
2-12
-------
Chapter 3
Cement Data
3.1 Data Requirements
Data requirements for U-ISIS module include sector-specific (in this case, cement-
specific) data as well as policy and economic parameters. The cement-specific data
requirements for U-ISIS cement module are discussed below.
As shown in Figure 3-1, the inputs are transmitted to the optimization part of the U-
ISIS model, where they are used to solve the selected BAU and policy cases. Potential
policy options may include cap-and-trade, emissions taxes, or emissions limits as
emission reduction mechanisms. After solving, the results are post-processed to
calculate values of various outputs of interest.
OTHERS (INPUTS)
Emissions & Products
Intensities, Emissions
Abatements
Others (elasticity,
discount & Inflation,
economic life etc.)
POLICY (INPUTS)
Emissions Reduction
/" \
Emissions Banking,
(COz, NO* etc.)
Allowance, Taxes etc.
INDUSTRY (INPUTS)
Products, Production
Capacity etc.
Production Cost (RMT,
VOM, FVC etc.)
Emissions Sources,
Intensities, Controls etc.
Fuel Types & Efficiencies
(Primary & Secondary)
Transportation &
Transport Cost
c>
-------
the selected base and policy cases. After solving, the results are post-processed to
calculate values of various outputs of interest. The output data are exported to Excel
spreadsheets for further analyses and graphical representation of selected results.
The U-ISIS model allows the user to select the policy case to be evaluated. In a cap-and-
trade policy scenario, separate caps on pollutants of interest can be specified based on
emissions from a selected reference year. The cap-and-trade policy scenario can be run
with or without banking of emissions and can include de minimis requirements.
Traditional rate-based policy scenarios with unit-specific emission reduction
requirements specified by the user can also be modeled in the U-ISIS model.
3.2 Cement-Specific Data
The inputs to the U-ISIS cement module can be broadly categorized into three main
components:
• Industry production, fuel, and emissions
• Control technologies, and emission abatement approaches
• Policy and economic parameters
3.2.1 Industry, Fuel, and Emissions
3.2.1.1 Existing, Planned/Committed, and Potential Units
The U-ISIS cement module contains information on 110 cement plants that were in
existence in 2010 (USGS, 2012). Additionally, the model also includes potential kiln
representations that may come on line as a result of endogenous capacity addition (i.e.,
new production capacity [by state] and replacements [by kilns]).
Some cement plants have multiple kilns. For example, in 2009 there were 189 kilns in
114 plants. Each kiln modeled in U-ISIS cement module is characterized by its location,
design (i.e., wet, diy, preheater, or precalciner), daily and annual clinker capacity3,
vintage, and retirement information when available (PCA, 2006). In addition, each kiln
is characterized by its average variable costs components (AVC).
3.2.1.2 Average Variable Costs
In previous economic analyses, five input variables in cement production have been
identified to determine the kiln-level AVC functions: raw materials, repair and
maintenance, labor, electricity, and fuel (Depro et al., 2007; Depro, 2010; Depro and
Lentz, 2010). Raw materials serve as the kiln feed, and repair and maintenance are
required for periodic upkeep of the kiln. Labor is used in the quarry, in the operation of
the kiln, and for packing. Electricity is consumed mainly by the auxiliary equipment
and fuel is largely consumed in the kilns. The AVC for raw materials, labor, repair and
3 Annual clinker capacity as reported by PCA refers to the kiln daily capacity multiplied by 365 less normal downtime days, where daily capacity is the
normal clinker oulpul a kiln can produce per day given a realistic work pauern and normal down lime clays are number of day? ofdowniime required for
maintenance, repair or cleanup.
3-2
-------
maintenance, and electricity was determined following the methodology in the EPA
regulatory impact analysis of the cement kiln dust rulemaking (EPA, 1998).
3.2.1.3 Cement Demand Centers
The U.S. cement markets are organized in state-specific demand centers. Figure 3-2
shows the distribution of Portland cement kilns operating in 2009. Each state
containing at least one kiln is shaded. The U-ISIS-cement module simulates each cement
plant's ability to compete in each of the demand centers as a function of the plant's
production cost and transportation cost associated with supply to each demand center.
3.2.1.4 Portland Cement Demand
One of the key data inputs for the U-ISIS-cement module is the demand projection for
each demand center. In general, this demand is a function of gross domestic product
(GDP) growth, interest rates, special construction projects (e.g., highways), and public
sector construction spending. Portland cement demand was 128 million metric tons in
2005 but only 71 million metric tons in 2010. PCA expects cement demand will reach
180 million metric tons by 2035. Should this happen it would reflect an increase of
nearly 110 million metric tons with a compound annual growth rate of 3.85 percent.
Cement demand through year 2035 is reported in the PCA Long-Term Cement
Consumption Outlook (PCA, 2009b). PCA projections of cement demand (in million
metric tons) by state through 2035, in 5-year increments, are shown in Table 3-2 (2005
data is given for comparison).
Portland Cement Kilns Locations
• Ce
111 St;
Source: EPA 2009 Data
Figure 3-2. Distribution of Cement Kilns in the United States as of 2009
3-3
-------
Table 3-2. Portland Cement Demand in Millions of Metric Tons (2009 projections)
State Demand Center
2005
2015
2020
2025
2030
2035
Alabama
1.92
2.02
2.16
2.32
2.49
2.67
Arizona
4.77
3.85
5.16
6.55
7.95
9.59
Arkansas
1.30
1.29
1.41
1.54
1.67
1.81
California
16.01
13.88
15.43
17.11
18.95
20.96
Colorado
2.55
2.76
3.08
3.47
3.90
4.33
Connecticut
0.82
0.77
0.83
0.88
0.94
0.98
Delaware
0.22
0.26
0.28
0.29
0.31
0.32
District of Columbia
0.21
0.18
0.18
0.17
0.18
0.18
Florida
12.28
8.99
11.05
13.47
16.33
19.67
Georgia
4.75
3.95
4.96
5.67
6.17
6.65
Hawaii
0.44
0.47
0.49
0.52
0.55
0.57
Idaho
0.70
0.70
0.82
0.92
1.03
1.15
Illinois
4.64
4.29
4.62
4.99
5.44
5.98
Indiana
2.27
2.30
2.49
2.73
2.98
3.25
Iowa
1.94
1.98
2.04
2.10
2.15
2.20
Kansas
1.55
1.65
1.71
1.78
1.85
1.91
Kentucky
1.60
1.48
1.59
1.71
1.84
1.97
Louisiana
2.23
2.91
3.15
3.29
3.42
3.52
Maine
0.35
0.29
0.31
0.32
0.34
0.35
Maryland
1.66
1.63
1.78
1.94
2.12
2.31
Massachusetts
1.26
1.11
1.21
1.31
1.43
1.56
Michigan
3.06
2.05
2.19
2.44
2.68
2.75
Minnesota
2.06
1.86
2.09
2.34
2.57
2.80
Mississippi
1.14
1.50
1.58
1.66
1.76
1.85
Missouri
2.87
2.71
3.18
3.58
3.90
4.29
Montana
0.38
0.42
0.45
0.48
0.51
0.53
Nebraska
1.37
1.46
1.55
1.66
1.76
1.87
Nevada
2.63
2.33
3.26
4.27
5.14
6.16
New Hampshire
0.37
0.33
0.36
0.39
0.42
0.45
New Jersey
2.06
1.89
1.99
2.09
2.18
2.28
New Mexico
0.91
0.94
1.00
1.07
1.12
1.15
New York
3.29
3.60
3.73
3.87
3.98
4.08
North Carolina
3.25
3.43
3.92
4.50
5.16
5.88
North Dakota
0.36
0.47
0.47
0.46
0.44
0.42
Ohio
4.06
3.61
3.75
3.90
4.02
4.12
Oklahoma
1.67
1.68
1.76
1.87
1.98
2.11
3-4
-------
State Demand Center
2005
2015
2020
2025
2030
2035
Oregon
1.24
1.22
1.40
1.61
1.86
2.15
Pennsylvania
3.44
3.49
3.64
3.78
3.89
3.97
Rhode Island
0.19
0.17
0.18
0.19
0.19
0.20
South Carolina
1.94
1.81
1.99
2.19
2.38
2.56
South Dakota
0.48
0.55
0.57
0.61
0.64
0.67
Tennessee
2.52
2.38
2.61
2.89
3.19
3.53
Texas
15.09
17.51
21.03
24.10
27.48
30.58
Utah
1.53
1.82
2.08
2.40
2.78
3.23
Vermont
0.16
0.16
0.17
0.18
0.20
0.21
Virginia
2.87
2.84
3.13
3.45
3.80
4.17
Washington
2.24
2.50
2.75
3.04
3.36
3.73
West Virginia
0.54
0.59
0.60
0.60
0.59
0.58
Wisconsin
2.37
2.18
2.32
2.45
2.59
2.74
Wyoming
0.47
0.55
0.55
0.55
0.55
0.54
Total
128.03
122.81
139.06
155.70
173.14
191.51
3.2.1.5 Transportation-Interregional Trade
In U-ISIS-cement module, a transportation matrix is used to describe the costs for
transporting cement from kiln and import district locations to demand centers. To
develop these costs, information on distances between supply and demand points and
costs of transportation modes (truck, rail, or water transport) was obtained. In
particular, the Transportation Routing Analysis Geographic Information System
(TRAGIS) model (TRAGIS, 2003) was used to develop the origin-destination distances as
described below. Also, in the matrix, the applicable lowest cost transportation option is
used to connect a supply point with a destination. While the cement demand centers
are interlinked through a transportation matrix, the competition is generally maintained
on a regional level because the cost of transporting cement is relatively high.
As mentioned above, the TRAGIS model was used to calculate transportation distances
associated with delivery of Portland cement from domestic U.S. manufacturing sites and
import terminals to each of the demand centers. Operation of the TRAGIS program
requires the identification of a shipment's origin and destinations, as well as the
selection of mode of transportation (highway, rail, or water) for each origin-destination
pair. The TRAGIS program does not use actual addresses or coordinates as origins or
destinations. Instead, the program uses a fixed set of pre-determined locations within
the model corresponding to locations in a number of U.S. cities and towns. The locations
of domestic Portland cement manufacturers used correspond to the lists of
manufacturing plants published in the Cement Americas (Penton Media Inc.), North
American Cement Directory (Cement Americas 2008); and the PCA, U.S. and Canadian
Portland Cement Industry Plant Information Summary December 31, 2006 (2007).
3-5
-------
The cost of transportation for interregional trade takes into account the impact of
terminals and the frequent use of modes other than truck. According to the USGS
(2007a), nearly half of cement shipments reach customers via terminals rather than
direct. Shipments to terminals are more than 80 percent by rail or water, rather than
by truck. Table 3-3 provides information on cost per ton mile for bulk shipping via truck,
rail, and barge. In U-ISIS-cement, transportation costs for each mode of transportation
(truck, rail, and barge) are calculated from each kiln and import district to each demand
center. For a given origin-destination pair, the dominant mode (lowest cost of shipment)
is used to determine the transportation cost for that route. The TRAGIS model was used
to arrive at the origin-destination distances and feasible routes for rail and barge; Google
Maps was used for truck transportation. The shipment costs shown in Table 3-3 were
used to determine the transportation costs for the routes in U-ISIS cement module.
Table 3-3. Bulk Shipment Costs (Cents per Metric Ton per Mile)
Mode
BTS1
WDOT2
NACT3
AU4
Average Used
in U-ISIS
Truck
33.06
7.64
11.02
7.18
14.73
Rail
2.78
3.32
5.51
N/A
3.87
Barge
0.89
1.05
2.20
N/A
1.38
1. BTS=Bureau of Transportation Statistics
2. WDOT=Washington State Department of Transportation
3. NACT= North American Cement Transportation
4. AU=American University
3.2.1.6 Imports
U.S. cement markets receive imported quantities of cement and clinker from a number
of countries, and these imports arrive at more than 30 import districts (USGS, 2007b).
In U-ISIS-cement module, international supplies from exporting countries to U.S. import
districts are modeled using supply elasticity and then these imports are transported to
the demand centers.
The five largest international suppliers of cement and clinker to the U.S. in 2005 were
China, Thailand, Venezuela, South Korea, and Greece. In 2010, the five largest
international suppliers were Canada, Korea, China, Mexico, and Colombia (USGS,
2012). However, an econometric study was conducted to provide an estimate of
international supply elasticity for supplies the top five international suppliers in 2005
and the rest of the world. The results of this study (Burtraw, 2010) reflected that the
best estimate of the international supply elasticity of cement and clinker from China,
Thailand, Venezuela, South Korea, Greece, and the rest of the world into the U.S. is
3.94.This value indicates that if the price of cement were to increase by 1 percent within
any import district in the U.S., then, ceteris paribus, the quantity of cement imported
from each of these five supply countries into that district would increase by 3.94
percent.
Table 3-4 shows the import levels by major USGS Customs District for 2010, estimated
by using USGS data (USGS, 2012).
3-6
-------
Table 3-4. Portland Cement and Clinker Imports in Million Metric Tons, by Major USGS
Customs District in 2010
USGS Customs District Quantity, million
metric tons
Alaska, Anchorage
0.11
California, Los Angeles
0.03
California, San Francisco
0.20
Florida, Miami
0.14
Georgia, Savannah
0.15
Hawaii, Honolulu
0.27
Louisiana, New Orleans
0.07
Maine, Portland
0.05
Massachusetts, Boston
0.02
Michigan, Detroit
0.93
Minnesota, Minneapolis;
0.01
Missouri, St. Louis
0.05
New York, New York
0.21
New York, Ogdensburg
0.16
North Carolina, Wilmington
0.11
Ohio, Cleveland
0.55
Oregon, Columbia-Snake
0.33
Pennsylvania, Philadelphia
0.14
Texas, El Paso
0.26
Texas, Houston-Galveston
0.64
U.S. Virgin Islands
0.02
Vermont, St. Albans
0.07
Virginia, Norfolk
0.11
Washington, Seattle
0.91
Total
6.62
Source: USGS, 2012 Table 18
3.2.1.7 Capacity Changes
Cement plants have a relatively long lifespan, typically 50 years or more (FLSmidth,
2007). Various factors, including (but not limited to) raw material availability in the
quarry, technology changes, productivity, efficiency, longevity, reliability, maintenance,
and long-term costs can affect the lifespan of a cement kiln/plant. In U-ISIS cement
module, retirements and projected retirements of existing kilns were based on
information from PCA on capacity expansion estimates. These estimates were
supplemented with information from individual cement companies on their plans for
shut-downs, new construction, and kiln consolidation (PCA, 2004). Further, as
mentioned earlier, U-ISIS-cement includes algorithms for endogenous capacity growth
3-7
-------
and retirement of kilns. To determine capital recovery factor for capital costs associated
with kiln capacity changes, an economic life of 25 years and an interest rate of 15
percent are used. Capital costs in 2005 $ per ton of clinker for new, replacement of wet,
and replacement of dry capacity are 208, 296, and 238, respectively (PCA, 2009b).
3.2.1.8 Fuel Intensity
The Annual Energy Outlook energy use profile for 2005 (EIA, 2008) is shown in Figure
3-3. In 2005, the cement sector consumed 451.2 trillion Btu (476.0 trillion kJ) of energy
(EIA, 2008). As shown in Figure 3-3, the primary fuel being burned in kilns is coal. Coal
is projected to remain the dominant fuel used by the U.S. cement industry. However,
there has been an increasing trend towards using other fuels, particularly alternative
fuels, such as coke,4 waste tires, and other wastes, especially oily wastes.
3.20%
0.50%
0.76%
Natural Gas
Fuel Oil
Coal
Petroleum Coke
Tires
Other Solid Wastes
Liquid Wastes
Figure 3-3. Commercial Fuel Use Profile by U.S. Cement Industry in 2005.
Source: EIA, 2008
In the U-ISIS-cement module, to determine the fuel intensity of each kiln, correlations of
kiln type to heat input and/or gas flow were developed (see Appendix A). Once
determined, the kiln's specific fuel intensity is used to calculate fuel cost for each kiln.
PCA's data on heat input to various kilns by type were used to develop kilns' fuel
intensities. The data, expressed in heat input per unit of clinker (specific fuel
consumption [SFC]) and exhaust gas flow rate (wet) (EGFW) per unit of clinker, are
summarized in Table 3-5.
PCA does not specify if "coke" is metallurgical coke or petroleum coke. Authors believe it is the latter.
3-8
-------
Table 3-5. Specific Fuel Consumption and Total Exhaust Gas Flow Rate (wet) for Various
Kiln Types
Kiln Type
SFO
EGFW
MMBtu/short ton
Nm3/kg
Wet
6.0
3.4
Dry
4.5
1.8
Preheater
3.8
1.5
Preheater/Precalciner
3.3
1.4
a. SFC=specific fuel consumption; (Source: EPA, 2007 (Table 3-3)
b. EGFW=exhaust gas flow rate (wet): (Source: PC A, 2004; original data in metric units)
For each individual kiln, the U-ISIS model determines the optimal fuel type(s) based on
the regional cost of the chosen fuel and the kiln's specific fuel intensity.
3.2.1.9 Emission Intensities
The design of the U-ISIS model can accommodate any number of pollutants of interest.
In U-ISIS-cement, each kiln is characterized by its NOx, SO2, PM, HC1, Hg, total
hydrocarbon (THC), and CO2 emission intensities. These emission intensities were
developed using available data (Andover Technology Partners, 2009a).
NOx emissions from cement kilns result primarily from the following combustion
process: oxidation of fuel nitrogen (fuel NOx) and the oxidation of nitrogen in the
combustion air (thermal NOx) • Oxidation of nitrogen in the feed materials (feed NOx) can
also influence total NOx emissions. Table 3-6 shows NOx emission intensities for cement
kilns in lb/ton of clinker and in lb/MMBtu (EPA, 2007).
Table 3-6. Estimated Uncontrolled NOx Emission Intensities for Cement Kilns
Heat Input,
Uncontrolled NOx Emissions
Kiln Type
MMBtu/ton of
clinker
lb/ton of clinker *
lb/MMBtu
Wet
6.0
9.7
1.62
Long Diy
4.5
8.6
1.91
Preheater
3.8
5.9
1.55
Preheater/Precalciner
3.3
3.8
1.15
* Average
Source: EPA, 2007 (Table 3-3 and Table 6-1)
SO2 emissions from cement kilns are the product of sulfur in the fuel as well as sulfur
in the feed materials. Sulfur in the fuel will oxidize to SO2 during pyroprocessing, and a
significant amount is likely to be captured in the form of sulfates as the gas passes
through the calcination zone. Compared to long diy and wet kilns, preheater and
preheater/precalciner kilns tend to be more effective at capturing fuel-generated SO2.
Accordingly, oxidation of sulfur in the feed materials is likely to be a major component
of total SO2 emissions. Table 3-7 shows average SO2 emissions for each kiln type for
each state (Andover Technology Partners, 2009a). State-specific emission intensities of
3-9
-------
S02 were determined from emissions reported for the kilns in that state. For any state
where the emission intensity was not available for a kiln-type, the national average
emission intensity was assigned to kilns in that state.
CO2 emissions from cement kilns result from limestone calcination and fuel combustion.
Appendix A explains how CO2 emission intensities from calcination and combustion are
calculated in the U-ISIS model. Calcination releases 0.52 tons of CO2 per ton of clinker
produced, while fuel-based CO2 emission factors range from 199.52 lb C02/MMBtu for
coal to 105.02 lb C02/MMBtu for natural gas (Andover Technology Partners, 2009b,
2010a, and 2010b). Table 3-8 shows approximate CO2 and water (H2O) produced from
combustion for the most important fuels for cement kilns.
Table 3-7. Average SO2 Emissions for Each Kiln Type in Each State
State
SO2 (lb/ton clinker)
Precalciner
Preheater
Dry
Wet
AL
0.09
0.61
9.02
13.99
AZ
1.30
0.07
8.51
7.81
AR
2.94
2.32
3.45
7.59
CA
0.33
1.12
0.81
0.00
CO
0.32
0.10
1.07
4.24
CT
1.15
2.32
9.02
7.81
DE
1.15
2.32
9.02
7.81
FL
0.45
0.02
9.02
7.81
GA
0.95
2.73
11.42
12.25
ID
0.13
2.32
9.02
0.70
IL
5.58
5.77
5.88
9.50
IN
1.15
2.32
9.02
7.81
IA
1.15
2.32
9.02
7.81
KS
1.25
5.88
24.85
8.61
KY
0.19
5.04
12.53
7.81
LA
1.15
2.32
9.03
7.81
ME
0.30
0.30
N/A
11.98
MD
3.14
3.95
7.31
7.49
MA
1.15
2.32
9.03
7.81
MI
5.23
2.32
5.31
18.00
MN
1.15
2.32
9.03
7.81
MS
0.09
0.61
9.02
13.99
MO
2.18
2.32
1.51
7.02
MT
0.13
2.32
9.02
0.70
NE
1.25
5.88
24.85
8.61
NV
0.49
0.05
1.51
7.81
NH
1.15
2.32
9.02
7.81
NJ
1.15
2.32
9.02
7.81
NM
1.30
0.07
8.51
7.81
3-10
-------
State
SO2 (lb/ton clinker)
Precalciner
Preheater
Dry
Wet
NY
0.30
0.30
9.02
11.98
NC
1.15
2.32
9.02
7.81
ND
1.15
2.32
9.02
7.81
OH
0.19
5.04
12.53
7.81
OK
1.25
5.88
24.85
8.61
OR
0.13
2.32
9.02
0.70
PA
1.15
2.32
9.02
7.81
RI
1.15
2.32
9.02
7.81
SC
0.95
2.73
11.42
12.25
SD
0.32
0.10
1.07
4.24
TN
0.95
2.73
11.42
12.25
TX
1.15
2.32
9.02
7.81
UT
0.13
2.32
9.02
0.70
VT
1.15
2.32
9.02
7.81
VA
0.95
2.73
11.42
12.25
WA
0.53
2.32
9.02
3.80
WV
3.14
3.95
7.31
7.49
WI
1.15
2.32
9.02
7.81
WY
0.32
0.10
1.07
4.24
National Average
1.15
2.32
9.02
7.81
N/A = Not available.
Table 3-8. Approximate CO2 and H2O Produced from Combustion of Fuels
Variable Name
Flue Gas
(per MMBtu)
Tires
Petroleum
Cokes
Heavy
Fuel
Oil
Rosemont
PRBa
Logan,
WV
BITb
Natural
Gas
LBMC02MMBTU
LBC02MMBTU
LBMH20MMBTU
LBH20MMBTU
lb moles C02 4.26 4.83 3.85 4.25 4.53 2.39
lb C02 187.44 212.56 169.32 186.83 199.52 105.02
lb moles H20 2.76 2.76 2.23 3.23 4.14 2.77
lbs H20 49.71 49.73 40.26 58.21 74.64 49.92
a PRB = Powder River Basin coal
b WV BIT = West Virginia bituminous coal
3-11
-------
3.2.2 Control Technologies and Emission Abatement Approaches
U-ISIS-cement contains information on abatement approaches for NOx, SO2, PM, HC1,
Hg, THC, and CO2 emissions described above. The three categories of abatement
approaches included are: process modifications and upgrades, raw material and/or fuel
substitution, and mitigation technologies. For each emission abatement approach,
where possible, information on the following parameters was developed (Andover
Technology Partners, 2009b and 2009c) and included in the model: capital cost, fixed
operating cost, variable operating cost, emission reduction performance for all of the
pollutants, impacts on fuel and/or raw material use, impact on electricity consumption,
byproduct generation and cost, and impact on water use.
To estimate capital recovery factors for capital costs associated with control
technologies, economic life values of 15 years and an interest rate of 7 percent are
generally used, but different values can be selected by the user. Payback periods and
technical life for the energy efficiency measures shown in Tables 3-14 through 3-17 (at
the end of this chapter) are given in Worrell and Galitsky (2004). Economic life for each
of these measures can be taken to be the average of the technical life and the payback
period. Again, an interest rate of 7 percent can be used for capital recovery in the
absence of more specific information.
Tables 3-9 through 3-12 show the NOx, SO2, CO2, HC1, Hg, and THC emissions control
technologies being used in the U-ISIS-cement module. Tables also reflect the impacts of
these technologies on pollution reduction, electricity use, and water use. Multimedia
impacts of changes to the capacity of cement kilns are listed in Table 3-13. Tables 3-14
through 3-17 show the electricity consumption and heat input changes accomplished
as a result of implementation of energy efficiency measures for raw materials
preparation, clinker making, finish grinding, and plant-wide measures, respectively.
More details on cost, efficiency, and co-impacts for each of the above control
technologies or process modifications are given in Appendix A.
3.2.3 Policy and Economic Parameters
3.2.3.1 Policy Parameters
The U-ISIS model framework allows the user to select a variety of potential policy options
for evaluation. The user can select from cap-and-trade policy (with or without de minimis
requirements), emissions charge policy, or rate-based policies. In a cap-and-trade policy
scenario, separate caps on pollutants of interest can be specified. The user has the
option to run a cap-and-trade policy scenario with or without banking of emissions.
Further, a cap-and-trade policy scenario can include de minimis requirements, where
the user defines a minimum level of emission reduction required for each emission unit.
As mentioned before, it is also possible for the user to input an emission charge for
pollutants of interest. Furthermore, traditional policy scenarios (rate-based policies)
with unit-specific emission reduction requirements specified by the user can be modeled
in U-ISIS.
The user can specify the policy horizon (time period) to be used for the model runs. Since
climate-related simulation horizons can be long (e.g., 40 years), the user may choose to
3-12
-------
run U-ISIS with blocks of years (e.g., 5-year blocks). The simulation horizon and blocks
of years can be chosen by the user subject to availability of data.
3.2.3.2 Additional Economic Parameters
In the U-ISIS framework, the following additional economic parameters are used:
discount rate, escalation rates, and demand elasticity. For the U-ISIS-cement module,
the default discount rate has been chosen as 7-percent, as recommended by the U.S.
Office of Management and Budget (OMB) for project evaluation (OMB, 1992). Escalation
rates used can be found in the "ISIS_Inputs.xls" workbook. Escalation rates for labor
are based on historical data from Bureau of Labor and Statistics (BLS) (BLS, 2008). Raw
materials escalation factors are calculated from historical price data of crushed
sandstone and gypsum from USGS. The escalation factors for variable operating and
maintenance costs are based on Chemical Engineering Index. Escalation factors for
various fuels and electricity are estimated based on data from the EIA's Annual Energy
Outlook 2008 (RTI, 2010). Demand for cement is relatively inelastic and an elasticity
value of -0.88 is used in the model (EPA, 1998).
3-13
-------
Table 3-9. NOx Control Technologies for Cement Kilns
Control Type
Impact on Emissions,
±% change
Electricity Consumption,
kWh/ton of cement
Water
Consumption,
gal/ton of
cement
NOx
SO2
FIVE
Hg
Other
Grinding
Kiln
Operation
Low NOx Burners -
Indirect Firing
-20% to -30%
No
impact2
03
-1.24
03
Mid Kiln
Firing-Tires
-20% to -40%
May vary5
03
06
03
Low NOx Burner + Mid
Kiln Firing- Tires
-20% to -40%
May vary5
03
-1.24'6
03
Low NOx Burners +
Tire Derived Fuel
-20% to -40%7
May vary5
03
-1.24'6
03
Low NOx Burner +
Selective Non Catalytic
Reduction
-50%?
No data5
No
data8
No data8
No
data8
0
-1.24'9
+ 1.259
Low NOx Burner +
Selective Catalytic
Reduction
-90%?
Oxidation
10
No
data8
Oxidation
10
No
data8
0
-1.24'9
+ 1.259
Low NOx Burners +
CemStara
-30% i
-1.3 (wet
process);
-1.9 (diy
process)12
-1.24'12 from
LNB and
-1.5 (wet
process) or
-2.2 (diy
process) from
CemStar13
0
CemStar/Fly Ash
Injection
-20%4
-1.3 (wet
process);
-1.9 (diy
process)13
-1.5 (wet
process) or
-2.2 (diy
process) from
CemStar13
0
Notes to Table 3-9 on following pages
3-14
-------
1. See EPA (2004), Table 5-1.
2. See Andover Technology Partners 2009b memo.
3. These technologies do not use and do not affect the consumption of water in raw mix preparation or significantly affect electric
power consumption in cement manufacturing processes.
4. Conversion from direct firing systems typical of wet and dry process kilns and older preheater kilns to indirect firing systems,
required to implement low NOx burner (LNB) technology, could result in reductions in primary air fan and kiln-induced draft fan
power requirements and concomitant slight increases in coal conveying power requirements. A reduction in fan/blower power on
the order of 100 hp might be anticipated for a moderately sized (300,000 - 500,000 clinker tons per year) kiln converted from
direct firing to indirect firing. No power savings on adding an LNB would be anticipated if the kiln system is already indirect-fired.
5. SO2 emissions from cement kilns are strongly related to fuel and raw materials sulfur content and to method of kiln operation.
Sulfur content of tire-derived fuel (TDF) (typically, 1.24% by weight, dry) may be higher or lower than the sulfur content of other
fuels commonly used in cement kilns, such as coal or coke. Therefore it is not practical to relate SO2 emissions to the use of these
NOx control methods. With respect to use of an SO2 wet scrubber using ground limestone, it is assumed that, for wet process kiln
systems, uncontrolled SO2 emissions of 8.2 lb per ton of clinker (8.9 lb/ton of cement) (EPA, 1995: Table 11.6-8) are treated by
use of a stoichiometric amount (with respect to uncontrolled SO2) of limestone of 90-percent purity in a 15-percent limestone
slurry. Limestone and water consumption are 15.5 lb and 87.7 lb, respectively, per ton of cement produced. For a precalciner
kiln, it is assumed that uncontrolled SO2 emissions of 1.1 lb per ton of clinker (1.2 lb/ton of cement) (EPA, 1995: Table 11.6-8)
are treated similarly (stoichiometric amount limestone of 90% purity in a 15% limestone slurry). Limestone and water consumption
are 2.1 lb and 11.8 lb per ton of cement produced.
6. Kiln system and raw materials grinding electric power consumption would not be significantly affected by introduction of tires or
tire-derived fuel.
7. Combined effect will vary. Low NOx 20 to 40 percent, mid kiln firing 20 to 40 percent.
8. While there may be theoretical or limited experimental bases to assume increases or decreases in emissions of various pollutants
in connection with NOx emission controls, statistics on such effects are not available. See, generally, EPA (2007a), Chapter 11.
9. Assumes selective non-catalytic reduction (SNCR) or selective catalytic reduction (SCR) both with stoichiometric addition of NH3
in 20-percent solution, typical precalciner NOx emissions of 4.2 lb/ton clinker before treatment (EPA 1995, Table 11.6-8), and
0.92 tons of clinker being used in each ton of cement. It should also be noted that SNCR is not applicable to wet process, long dry
process, and many preheater kilns because the kiln gas exit temperatures are too low from those units for the necessary reactions
to take place. See EPA (2007). SNCR is assumed to achieve 63-percent reduction in NOx emissions from applicable kiln systems.
See EPA (2007), Chapter 8. There may be an attendant small increase in kiln system electric power in connection with injection
of water into the kiln system due to an increase in gas volume handled by the kiln's fan system. The increase is not considered in
the calculation.
10. Typically, up to about 1 percent of SO2 can be oxidized to SO3. Elemental form of Hg oxidized across SCR provided there is sufficient
concentration of halogens in flue gas.
11. On alb / ton of clinker basis.
12. Use of the CemStar process may have multiple effects on electric power consumption, including reduced raw mix preparation
costs in the event that unground CemStar material is fed to the rotary kiln, and reduced fan power requirements resulting from
3-15
-------
reduced kiln gas volumes in connection with both combustion gases and raw mix calcination. Assuming a 5-percent replacement
of raw mix with carbonate-free CemStar material, a concomitant 5-percent reduction in raw mix preparation energy is also
assumed. An approximate 5-percent reduction in kiln electrical energy consumption is assumed based on a roughly 5-percent
reduction in kiln exit gases from both calcination and combustion.
13. CemStar and fly ash injection are expected to have similar effects on raw mix preparation and kiln process electrical energy
requirements.
14. See NESCAUM (2000) for effect of CemStar on NOx reductions from cement kilns. Investigation of the combined effects of multiple
technologies on pollutant emissions was not carried out for this summary.
General Notes
A. Cement kiln processes do not use steam.
B. Cement kiln dust (CKD) disposal rates as functions of NOx or SO2 emissions control technology has not been reported. Typical
CKD disposal rates range from 0.042 to 0.115 tons CKD/ton clinker (0.046 to 0.125 tons CKD/ton cement assuming 0.92 tons
clinker per ton of cement).
C. CKD disposal costs vary widely by region, CKD characteristics, CKD volumes, and location of disposal site (on-site or off-site). In
addition, disposal costs are expected to be of comparable cost to transportation costs in connection with off-site disposal. For
example, prices at three northern California landfills range from $26/ton to $69/ton of CKD plus transportation, which may range
from approximately $500 to $1,200 per truck load. Pers. Comm., E. Learner (ARCADIS) Feb. 26, 2009. CKD disposal prices would
be expected to be generally unaffected by the type of cement kiln pollution controls employed.
References for Table 3-9 (references called out in the above notes can be found at the end of the chapter)
Air Pollution Controls and Efficiency Improvement Measures for Cement Kilns, Prepared by ARCADIS Under Contract No. EP-C-04-
023, March 31, 2008.
Srivastava et al., ES&T, March 2006, pp. 1385-1393.
Theoretical Approach for Enhanced Mass Transfer Effects in-Duct Flue Gas Desulfurization Processes. Final Report. 1992.
3-16
-------
Table 3-10. SO2 Control Technologies for Cement Kilns
Control Type
Impact on Emissions,
±% change
Electricity Consumption,
kWh/ton of cement
Water
Consumption,
gal/ton of cement
NOx
S02
PM
Hg
Other
Grinding
Kiln
Operation
Wet Scrubber
-90% to -
95%
-80%
-50% for THC
-99.9% for
HC1
+0.2
(wet process
kilns)2
+52
+ 12 (wet process);
+ 1.5 (diy process)3
Diy Lime
Injection
-50%*
-75% for HC1
5
5
5
Notes to Table 3-10
1. See Andover Technology Partners 2009b memo.
2. Electric power consumption in connection with wet scrubbing is assumed to be primarily a result of: 1) grinding of limestone to
below 45 to 74 micrometers in particle size and pumping limestone slurry; and 2) increased kiln exhaust gas fan power
requirements. The latter energy increase is caused by increased pressure drop demand on kiln ID fans due to conveying kiln exit
gases across the scrubber spray tower plus increased gas volume demand on kiln ID fans due to the addition of water in the
limestone slurry. Slurry preparation and pumping is assumed to require 20 kWh per ton of limestone. The spray tower pressure
drop is assumed to add 500 hp to ID fan requirements on a 300,000 to 500,000 ton per year cement kiln process.
3. SO2 emissions from cement kilns are strongly related to fuel and raw materials sulfur content and to method of kiln operation.
Sulfur content of TDF (typically 1.24% by weight, dry) may be higher or lower than the sulfur content of other fuels commonly
used in cement kilns, such as coal or coke. Therefore it is not practical to relate SO2 emissions to the use of these NOx control
methods. With respect to use of an SO2 wet scrubber using ground limestone, it is assumed that, for wet process kiln systems,
uncontrolled SO2 emissions of 8.2 lb per ton of clinker (8.9 lb/ton of cement) (EPA, 1995: Table 11.6-8) are treated by use of a
stoichiometric amount (with respect to uncontrolled SO2) of limestone of 90-percent purity in a 15-percent limestone slurry.
Limestone and water consumption are 15.5 lb and 87.7 lb, respectively, per ton of cement produced. For a precalciner kiln, it is
assumed that uncontrolled SO2 emissions of 1.1 lb per ton of clinker (1.2 lb/ton of cement) (EPA, 1995: Table 11.6-8) are treated
similarly (stoichiometric amount limestone of 90% purity in a 15% limestone slurry). Limestone and water consumption are 2.1
lb and 11.8 lb per ton of cement produced.
4. 20 percent at Ca:S stoichiometry of 2:1 to 5:1. See Dry Sorbent Injection report (ARCADIS, 1992).
5. Dry lime injection technology is not well developed for the cement industry and no statistics are available.
References for Table 3-10 (references called out in the above notes can be found at the end of the chapter)
Andover Technology Partners, Cost and Performance of Controls, March 10, 2009.
Srivastava et al., ES&T, March 2006, pp. 1385-1393.
Theoretical Approach for Enhanced Mass Transfer Effects in-Duct Flue Gas Desulfurization Processes. Final Report. ARCADIS, 1992.
3-17
-------
EPA (1995). Compilation of Air Pollution Emission Factors - Volume 1: Stationary Point and Area Sources. Fifth Edition, Supplements
A-F, AP-42. U.S. Environmental Protection Agency, Research Triangle Park, North Carolina.
Table 3-11. CO2 Control Technologies for Cement Kilns
Control Type
Impact on Emissions,
+/- % change
Electricity
Consumption,
Process Water
Consumption,
Cooling Water
Consumption,
ton/ton of clinker
NOx
SO2 PM Hg CO2 Other
MWh/ton of
clinker
ton/ton of
clinker
Solvent CCS1 New
-99%
-99% -85%
-0.022
0.429
4.82
Solvent CCS*
Retrofit
-99%
-99% -85%
-0.103
0.160
4.82
Oxy-combustion
-99%
-99% -85%
0.174
14.3
Notes to Table 3-11
1. CCS = carbon capture and sequestration.
References for Table 3-11
Andover Technology Partners, Cost and Performance of Controls, March 10, 2009.
EPA, Report to Congress on Cement Kiln Dust, Chapter 3: CKD Generation and Characteristics; U.S. Environmental Protection
Agency: 1993.
3-18
-------
Table 3-12. HC1, Hg, and THC Control Technologies for Cement Kilns
Control Type1
Impact on Emissions, ± % change
NOx
SO2
PM
Hg
Other
ACI2
-99.9%
-90%
-50% for THC
Membrane Bag
-99.9%
RTO3
-98% for THC
Diy Lime Injection
-50%
-75% for HC1
Wet Scrubber
-90%
to
-95%
-80%
Notes to Table 3-12
1. Feasible combinations of the above control technologies may be utilized as necessary.
2. ACI = activated carbon injection.
3. RTO = regenerative thermal oxidizer
Table 3-13.Multimedia Impacts of Process Capacity Replacement on Cement Kiln
Operation1
Water
Consumption,
gal/ton of
cement2
Electricity
Consumption,
kWh/ton of cement
Waste
Kiln Type
Grinding3
Kiln
Operation4
Generation
Rate, ton/ton
of cement5
Disposal
Cost,
$/ton of
cement
Wet to
Precalciner
-2142,6
-12s
-78
-0.072
see notes 9,10
Long Diy
to
Precalciner
No change2
Not
available7
Not
available7
-0.062
see notes 9,10
Preheater
Not
available7
Not
available7
to
Precalciner
No change2
on
see notes 9,10
Notes to Table 3-13.
1. Unless specifically noted otherwise, impacts are presented on a per-short-ton-of-cement
basis and generally make use of published data reported on a per-short-ton-of-clinker basis.
These values are converted to a per-ton-of-cement basis by assuming that cement consists
of 92-percent clinker. All weight units in this table are short tons.
2. Water is not normally consumed in dry process cement kiln processes, except on an
emergency basis to prevent damage to process equipment by hot gas or solid process streams.
Water is used in non-contact cooling processes that are common in both wet process and dry
process cement plants, some of which may or may not use closed circuit cooling systems
with no evaporation losses. Water is consumed in direct contact cooling processes such as
cement grinding in both wet process and dry process plants. However, the nature and amount
of such water consumption is not intrinsically different between wet process and dry process
plants.
3. Power data reported in Worrell and Galitsky (2004) are used for raw mix preparation and kiln
process electrical energy consumption for wet and all dry processes.
3-19
-------
4. Air pollution control device contributions to electric power consumption data on cement kiln
process systems are assumed to include only existing particulate matter control devices and
not scrubbers for SO2 or other criteria pollutants.
5. Solid waste from all process types is assumed to be CKD. Data used here are as reported in
EPA (1993).
6. Cement is assumed to consist of 92-percent clinker. Clinker is assumed to require 1.42 tons
of raw kiln feed (dry) per ton of clinker. Kiln feed loss on ignition is assumed to be 35 percent.
Kiln feed slurry is assumed to contain 36-percent water.
7. Industry-wide statistics on electrical energy use do not distinguish among the various cement
processing stages between the various dry processes - all dry process plants are averaged
together. Published data that distinguish between various process types combine all process
phases without indicating energy use for individual process phases. Therefore, electrical
energy use is reported here to be the same for all dry process plants. See Worrell and Galitsky
(2004).
8. As stated in note 7, industry data on electric power use in raw materials preparation and kiln
processing are not available for every dry process. Data are for all dry processes are
combined.
9. Data reported in EPA (1993) on CKG generation by cement plants doe not distinguish
between preheater and precalciner kilns with respect to net CKD generation rates.
10. CKD disposal prices vary widely by location. Transportation of CKD is a significant
component of the cost of disposal if disposal is off-site. For example, an estimated off-site
disposal cost ranging from $26 to $69 per ton of CKD for disposal, plus transportation costs
ranging from $500 - $1,500 per truck load for one northern California location, depending
on the landfill chosen. (ARCADIS, 2009) On-site disposal costs would include costs of
transportation, dust control, and landfill operation. These costs have not been determined
and would most certainly vary widely by location based on terrain, site geology, landfill
operating requirements, and permit and future closure costs.
11. Modern preheater cement kilns and precalciner cement kilns generate similar volumes of
CKD. Older preheater cement kilns are similar in CKD generation to long dry process cement
kilns.
Table 3-14.Energy Efficiency Measures for Raw Materials Preparation
Energy Efficiency Improvement Method
Electricity Consumption Change,
kWh/ton clinker
Dry
Wet
Pre-
heater
Pre-
calciner
ETS (Efficient Transport System)
-3.20
-3.20
-3.20
RMB (Raw Materials Blending)
-2.70
-2.70
-2.70
PCVM (Process Control Vertical Mill)
-0.90
-0.90
-0.90
HERM (High Efficiency Roller Mill)
-11.05
-11.05
-11.05
SBH (Shiny Blending and
-0.35
Homogenization)
WMCCC (Wash Mills with Closed Circuit
-12.00
Classifier)
RMTHEC (Raw Materials Transport High-
-5.05
-5.05
-5.05
-5.05
Efficiency Classifiers)
3-20
-------
Table 3-15.Energy Efficiency Clinker Making Measures
Energy Efficiency
Improvement
Method
Electricity
Change, kV
Consumption
/h/ton clinker
II
Heat Input Change,
ilMBtu/ton of clinker
Dry
Wet
Pre-
heater
Pre-
calciner
Dry
Wet
Pre-
heater
Pre-
calciner
EMCS
(Energy Management
and Control System)
-1.90
-1.50
-1.90
-1.90
-0.15
-0.21
-0.15
-0.15
SR
(Seal Replacement)
-0.02
-0.02
-0.02
-0.02
CSI
(Combustion System
Improvement)
-0.25
-0.35
-0.25
-0.25
IF
(Indirect Firing)
-0.16
-0.16
-0.16
-0.16
SHLR
(Shell Heat Loss
Reduction)
-0.20
-0.20
-0.20
-0.20
OGR
(Optimize Grate
Cooler)
0.90
0.90
0.90
-0.09
-0.10
-0.09
-0.09
CGC
(Convert to
reciprocating grate
cooler)
2.40
2.40
2.40
2.40
-0.23
-0.24
-0.23
-0.23
HRPG
(Heat Recovery for
Power Generation)
-18.0
EMD
(Efficient Mill Drives)
-2.00
-1.70
-2.00
-2.00
Table 3-16. Energy Efficiency Measures for Finish Grinding
Electricity Consumption Change,
Energy Efficiency Improvement kWh/ton clinker
Method
Dry
Wet
Preheater
Precalciner
EMPC (Energy Management and
-1.60
-1.60
-1.60
-1.60
Process Control)
IGMBM (Improved Grinding Media
-1.80
-1.80
-1.80
-1.80
[Ball Mills])
HPRP (High-Pressure Roller Press)
-16.00
-16.00
-16.00
-16.00
HEC (High-Efficiency Classifiers)
-3.85
-3.55
-3.85
-3.85
3-21
-------
Table 3-17.Energy Efficiency Plant-wide Measures
Electricity Consumption Change,
Energy Efficiency Improvement kWh/ton clinker
Dry Wet Preheater Precalciner
PM* (Preventive Maintenance)
-2.50
-2.50
-2.50
-2.50
HEM (High Efficiency Motors)
-2.50
-2.50
-2.50
-2.50
ASD (Adjustable Speed Drives)
-6.25
-6.00
-6.25
-6.25
OCAS (Optimization of Compressed
-1.00
-2.50
-1.00
-1.00
Air Systems)
*This is the only occurrence wherein "PM" does not stand for "particulate matter"
3-22
-------
3.3 References for Chapter 3
American University (2008). TED Case Studies: Cemex Case.
http://www.american.edu/TED/cemex.htia, accessed October 21, 2008.
Andover Technologies Partners (2009a). Memorandum: NOx, SO2 and CO2 Emissions
from Cement Kilns (Emissions Memo), from Jim Staudt to Ravi Srivastava,
Samudra Vijay, Elineth Torres. Dated March 10, 2009 (see Appendix A of this
document).
Andover Technology Partners (2009b). Memorandum: Costs and Performance Controls,
from Jim Staudt to Ravi Srivastava, Samudra Vijay, Elineth Torres. Dated March
10, 2009 (see Appendix A of this document).
Andover Technology Partners (2009c). Memorandum: GHG Mitigation Methods for
Cement, from Jim Staudt to Ravi Srivastava, Nick Hutson, Samudra Vijay,
Elineth Torres. Dated July 10, 2009 (see Appendix A of this document).
Andover Technologies Partners (2010a). Memorandum: Wet Scrubber Cost Algorithms,
from Jim Staudt to Ravi Srivastava, Elineth Torres, Keith Barnett. Dated
February 26, 2010 (see Appendix A of this document).
Andover Technologies Partners (2010b). Memorandum: Electrical Load for Wet
Scrubbers, from Jim Staudt to Elineth Torres. Dated May 7, 2010 (see Appendix
A of this document).
ARCADIS (1992). Theoretical Approach for Enhanced Mass Transfer Effects in-Duct
Flue Gas Desulfurization Processes. Final Report. ARCADIS, 1992.
ARCADIS (2009). Personal Comm. with E. Learner of ARCADIS, 26 February, 2009.
BLS (2008). Bureau of Labor Statistics, Databases, Tables & Calculators by Subject,
Major Sector Productivity and Costs Index. Series ID: PRS30006112, Sector:
Manufacturing.
http://data.bls.gov/PDO/servlet/SurveyOutputServlet9data tool=latest number
s&series id=PRS30006112, accessed November 24, 2008.
Burtraw, D. (2010). Supply Elasticity Estimation, memorandum to Ravi Srivastava, U.S.
EPA, March 10, 2010.
Cement Americas (2008). North American Cement Directory. Penton Media Inc.
Depro, Brooks, Rogozhin, Alex, and Wood, Dallas (2007). RTI International.
"Documentation for Portland Cement Kiln Cost Functions (2005)", Memorandum to
Keith Barnett, U.S. EPA, August 31, 2007 (see Appendix A of this document).
Depro, Brooks. (2010). RTI International. "ISIS Cement Production Costs",
Memorandum to Elineth Torres, U.S. EPA, March 31, 2010 (see Appendix A of
this document).
3-23
-------
Depro, Brooks and Lentz, Anthony. (2010). RTI International. "ISIS Cement Production
Costs", Excel Spreadsheet Attachment to Memo of Same Name from Brooks
Depro (2010), February 16, 2010.
EIA (2008) Annual Energy Outlook 2008, DOE/EIA-0383(2008), June 2008.
http:/ /www.eia. doe.gov/oiaf/aeo/pdf/0383(2008).pdf, accessed October 21,
2008.
EPA (1993). Report to Congress on Cement Kiln Dust, Chapter 3: CKD Generation and
Characteristics; U.S. Environmental Protection Agency: 1993.
EPA (1995). Compilation of Air Pollution Emission Factors - Volume 1: Stationary Point
and Area Sources. Fifth Edition, Supplements A-F, AP-42. U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina.
EPA (1998). June 1998. Regulatory Impact Analysis of Cement Kiln Dust Rulemaking.
Washington, DC: U.S. Environmental Protection Agency.
http://www.epa.gov/osw/nonhaz/industrial/special/ckd/ckd/ckdcostt.pdf,
accessed October 21, 2008.
EPA (2004). Alternative Control Techniques Document - NOx Emissions from Cement
Manufacturing, dated March 1994 (EPA-453/R-94-004)
EPA (2007). November 2007. Alternative Control Techniques Document Update - NOx
Emissions from New Cement Kilns. EPA-453/R-07-006. Research Triangle Park,
NC. U.S. Environmental Protection Agency.
FLSmidth & Co. A/S (2007). Q2 Report 2007. August 2007
http: //hugin. info/2106/R/1148414/219358.pdf, accessed October 21, 2008.
NESCAUM (2000). Status Report on NOx Controls for Gas Turbines, Cement Kilns,
Industrial Boilers, Internal Combustion Engines: Technologies & Cost
Effectiveness. Northeast States for Coordinated Air Use Management, December
2000. Accessed at http:/ /www.nescaum.org/documents/nox-2000.pdf/
OMB (1992). Guidelines and Discount Rates for Benefit-Cost Analysis of Federal
Programs. OMB Circular No.A-94 (Revised). U.S. Office of Management and
Budget, October 29, 1992.
http://www.whitehouse.gOv/omb/circulars/a094/a094.html#8, accessed
October 21, 2008.
PCA (2004). Innovations in Portland Cement Manufacturing. Portland Cement
Association. Edited by J. I. Bhatty, F. M. Miller, and S. H. Kosmatka. 2004.
PCA (2006). U.S. and Canadian Portland Cement Industry: Plant Information Summary.
Portland Cement Association, Skokie, IL, 2006.
PCA (2007). U.S. and Canadian Portland Cement Industry Plant Information Summary
December 31, 2006. Portland Cement Association, Skokie, IL, 2007.
3-24
-------
PCA (2009a). PCA Capacity Report. Flash Report. Portland Cement Association's
Economic Research Department. Updated October 19, 2009.
PCA (2009b). Forecast Report: Long-Term Cement Consumption Outlook. By Ed
Sullivan. October 28, 2009. Portland Cement Association.
RTI (2010). ISIS Input Price Escalation Factors, Excel Spreadsheet.
TRAGIS (2003). Transportation Routing Analysis Geographic Information System
(TRAGIS) User's Manual, Revision 0, Oak Ridge National Laboratory, June 2003.
https: / /tragis. ornl.gov/TRAGISmanual.pdf.
USGS (2007a). 2005 Minerals Yearbook: Cement. U.S. Geological Survey, p. 16.2,
February 2007,
http://minerals.usgs.gov/minerals/pubs/commoditv/cement/cemenmvb05.pd
f, accessed October 21, 2008
USGS (2007b). Mineral Commodity Summaries: Cement, U.S. Geological Survey, pp.
40-41, Januaiy 2007.
http://minerals.er.usgs.gov/minerals/pubs/commodity/cement/cemenmcs07.
pdf, accessed October 21, 2008.
USGS (2012). 2010 Minerals Yearbook: Cement U.S. Geological Survey, July 2012.
Worrell, E.; Galitsky, C. (2004). Energy Efficiency Improvement Opportunities for
Cement Making; Lawrence Berkeley National Laboratory, LBNL-54036, Jan
2004. An ENERGY STAR® Guide for Energy and Plant Managers,
http://www.osti.gov/energycitations/servlets/purl/821915-Re2kcK/native/.
3-25
-------
Chapter 4
Model Calibration
Large techno-economic models of U-ISIS framework size require model calibration as
they utilize an extensive amount of data which comes from different sources. This
chapter outlines calibration methodology that was used, discusses data used for
calibration, presents calibration results, and gives farther recommendations.
4.1 Calibration Methodology
Calibration methodology utilizes the concept of a calibration constant. The calibration
constant has been developed to account for possible errors in costs. The value of the
calibration constant, calconstfi), is set by trial and error during calibration. The objective
of the trial and error approach is to minimize the absolute difference in the reported and
module-predicted prices (which are marginal values of the supply equation) for each
USGS district.
In the first step of calibration, the module is set to run for 2005-2007 by making
appropriate changes in the input "Policy" worksheet, and GAMS input files. The import
quantities are then adjusted equal to reported import quantity for each of the import
district, except for those of Mexico and Canada.
In the next step, the impact of changing the calibration constant is monitored. This
impact of the calibration constant is assessed on estimated production quantities to
keep the difference between reported and module predicted production values within
reasonable limits5. The calibration constant modifies each kiln's variable cost of
production. The "Calibration" worksheet within the Inputs workbook has values of
calibration constant assigned for each USGS district. Finally, in the input GAMS file,
the values are assigned for a given USGS district to each of the kilns located in that
USGS district.
The module is first calibrated for year 2005, to obtain values of the calibration parameter
calconst(i) for the year. Next, calconst(i) values for 2005 are used as a starting point to
obtain values for the same parameter for 2006. Similarly, the process is repeated to
obtain the values for 2007. Then, an average of the calconstfi) values over the three years
is taken and used for the future module runs. Current values of the parameter
Calconst(i) being used in the module runs can be found in the worksheet "Calibration"
of the "ISIS_Inputs.xls" workbook.
5 There is no standard method to guide the user in determining an acceptable level of "error" in the reported and predicted
values for the purpose of calibration. In this works, we have set an acceptable level for the absolute gap between the
individual reported and predicted values to ±15%, although effort has been made to keep this level below 10% for most
of the quantities. However, due to discontinuities in the transportation matrix, error in the reported data, or other
unknowns, the gap in the estimated and reported values may be higher in certain markets.
4-1
-------
Calibration is a dynamic process, and it is recommended that module calibration be
performed periodically. In this fashion, any available new production, imports, or price
data could be utilized in the module.
4.2 Data for Calibration
Production quantities in various USGS production districts, import levels in the import
districts and reported cement prices for USGS districts are the key quantities used for
calibration of the module. Reported data for years 2005, 2006 and 2007 were used to
calibrate the module and obtain values of appropriate calibration parameters.
4.2.1 Cement Prices in USGS Districts
Reported cement prices in various USGS districts are shown in Table 4-1 for years 2005
through 2007. For a given year, the reported cement price shows significant variation
across USGS districts. For most districts the cement price has increased over the three
year period. USGS districts assignation is the location of the reporting production
facilities.
Table 4-1. Cement Prices ($/ton) for USGS Districts
USGS District
2005
2006
2007
Alabama
74.61
81.74
87.09
Illinois
80.29
89.81
91.17
Indiana
72.03
80.04
79.99
Kansas
76.58
86.50
92.87
Maryland
74.76
81.19
80.18
Missouri
78.92
86.61
89.36
Ohio
82.08
90.24
91.47
South Carolina
68.57
80.46
87.54
Maine and New York
80.74
92.53
96.62
Pennsylvania
79.73
89.99
90.85
Michigan and Wisconsin
79.83
84.82
90.12
Iowa, Nebraska and South Dakota
78.72
89.83
93.08
Florida
77.56
90.44
95.55
Georgia, Virginia and West Virginia
84.49
95.93
93.60
Kentucky, Mississippi and
84.37
89.81
88.90
Tennessee
Arkansas and Oklahoma
89.00
96.99
97.74
Texas
74.72
85.10
88.33
Arizona and New Mexico
75.76
84.16
86.64
Colorado and Wyoming
82.24
92.90
92.73
Idaho, Montana, Nevada and Utah
83.61
90.26
95.25
California
88.20
99.09
100.24
Oregon and Washington
79.38
90.72
90.72
4-2
-------
4.2.2 Cement Production in USGS Districts
Reported cement production in various USGS districts is shown in Table 4-2 for years
2005 through 2007. For a given year, the reported cement production shows significant
variation across USGS districts. Generally production dropped from 2005 to 2007, due
to a decrease in demand resulting from the economic downturn.
Table 4-2. Cement Production (tons) for Various USGS Districts
USGS District
2005
2006
2007
Alabama
5,647,141
5,733,121
5,578,650
Illinois
3,568,182
3,425,984
3,434,254
Indiana
3,370,868
3,334,492
3,285,999
Kansas
3,182,373
3,310,241
3,039,164
Maryland
2,813,098
2,922,227
3,305,160
Missouri
5,877,524
5,776,111
5,763,479
Ohio
1,086,879
1,064,833
1,010,077
South Carolina
3,601,251
3,654,162
4,057,414
Maine and New York
3,572,591
3,699,357
3,471,388
Pennsylvania
6,931,334
6,631,505
6,239,551
Michigan and Wisconsin
6,171,841
5,993,267
6,047,588
Iowa, Nebraska and South Dakota
4,962,606
5,024,335
4,889,607
Florida
6,311,835
6,477,181
6,076,137
Georgia, Virginia and West Virginia
2,612,478
2,696,253
2,528,648
Kentucky, Mississippi and
3,649,753
3,849,271
3,770,250
Tennessee
Arkansas and Oklahoma
3,097,495
2,979,547
2,879,982
Texas
12,737,207
12,510,131
12,038,919
Arizona and New Mexico
3,073,244
2,809,792
2,902,131
Colorado and Wyoming
2,918,920
2,842,861
2,797,240
Idaho, Montana, Nevada and Utah
3,400,630
3,354,333
3,308,645
California
12,747,128
12,069,207
11,941,198
Oregon and Washington
2,175,963
2,101,005
2,103,481
4.2.3 Cement Imports by Import Districts
Reported cement import quantity in various import districts is shown in Table 4-3 for
years 2005 through 2007. For a given year, the reported cement imports show a
significant variation across import districts. Generally imports dropped from 2005 to
2007, due to a decrease in demand resulting from the economic downturn. New Orleans
(LA), San Francisco (CA) and Tampa (FL) suffered large drops in the quantities of cement
imported in 2007 compared to that in 2005.
4-3
-------
Table 4-3. Cement Imports (except from Canada and Mexico) for Import Districts (tons)
Import District
2005
2006
2007
Baltimore, MD
146,300
203,500
18,700
Boston, MA
145,200
50,600
3,300
Buffalo, NY
6,600
4,400
0
Charleston, SC
1,212,200
1,100,000
363,000
Chicago, IL
1,100
1,100
0
Cleveland, OH
0
1,100
0
Columbia-Snake, ID-OR-WA
831,600
1,115,400
1,180,300
Detroit, MI
59,400
0
0
Duluth, MN
0
0
0
El Paso, TX
0
0
0
Great Falls, MT
0
0
0
Houston-Galveston, TX
2,880,900
3,705,900
3,641,000
Laredo, TX
0
0
0
Los Angeles, CA
3,357,200
3,759,800
2,029,500
Miami, FL
2,395,800
2,311,100
980,100
Minneapolis, MN
0
0
0
Mobile, AL
565,400
573,100
0
New Orleans, LA
4,503,400
5,090,800
1,171,500
New York, NY
1,390,400
1,327,700
810,700
Nogales, AZ
0
0
0
Norfolk, VA
767,800
793,100
447,700
Ogdensburg, NY
0
0
0
Pembina, ND
0
0
0
Philadelphia, PA
544,500
665,500
342,100
Portland, ME
0
0
0
Providence, RI
814,000
680,900
509,300
San Diego, CA
620,400
708,400
427,900
San Francisco, CA
2,599,300
3,081,100
1,524,600
Savannah, GA
88,000
204,600
376,200
Seattle, WA
368,500
733,700
644,600
St. Albans, VT
0
0
0
St. Louis, MO
0
0
3,300
Tampa, FL
3,825,800
3,791,700
1,523,500
Wilmington, NC
427,900
416,900
284,900
4.3 Results of Calibration
Reported and calculated values of cement prices by USGS district are shown for years
2005, 2006, and 2007 in Table 4-4, Table 4-5, and Table 4-6, respectively. As can be
seen from these tables, in most of the districts reported and calculated cement prices
are within 10 percent range, whereas in a few they are higher. It should be noted that
in general the price differentials are smaller in the year 2005 and have increased in
2007. In some USGS districts the reported and calculated prices are different than the
module predicted prices due to several factors, including error in reported prices,
4-4
-------
discontinuities in the transportation matrix, and/or unique factors for some specific
markets.
Table 4-4. Reported and Calculated Cement Prices in USGS Districts (2005)
USGS District
Reported
Calculated
%A
Alabama
74.61
78
5
Illinois
80.29
83
3
Indiana
72.03
84
16
Kansas
76.58
81
6
Maryland
74.76
74
-1
Missouri
78.92
82
4
Ohio
82.08
83
1
South Carolina
68.57
76
10
Maine and New York
80.74
73
-9
Pennsylvania
79.73
77
-3
Michigan and Wisconsin
79.83
81
2
Iowa, Nebraska and South
78.72
81
Q
Dakota
O
Florida
77.56
75
-3
Georgia, Virginia and West
Virginia
84.49
80
-6
Kentucky, Mississippi and
84.37
81
-4
Tennessee
Arkansas and Oklahoma
89.00
87
-2
Texas
74.72
78
4
Arizona and New Mexico
75.76
105
39
Colorado and Wyoming
82.24
98
20
Idaho, Montana, Nevada and
83.61
70
-16
Utah
California
88.20
85
-4
Oregon and Washington
79.38
69
-13
4-5
-------
Table 4-5. Reported and Calculated Cement Prices in USGS Districts (2006)
USGS District
Reported
Calculated
%A
Alabama
81.74
85
4
Illinois
89.81
83
-8
Indiana
80.04
85
6
Kansas
86.50
83
-4
Maryland
81.19
80
-1
Missouri
86.61
81
-6
Ohio
90.24
83
-8
South Carolina
80.46
85
6
Maine and New York
92.53
78
-16
Pennsylvania
89.99
80
-11
Michigan and Wisconsin
84.82
80
-6
Iowa, Nebraska and South
89.83
81
-10
Dakota
Florida
90.44
82
-10
Georgia, Virginia and West
95.93
86
-10
Virginia
Kentucky, Mississippi and
89.81
84
-6
Tennessee
Arkansas and Oklahoma
96.99
91
-6
Texas
85.10
81
-4
Arizona and New Mexico
84.16
105
25
Colorado and Wyoming
92.90
100
8
Idaho, Montana, Nevada and
90.26
83
-8
Utah
California
99.09
85
-14
Oregon and Washington
90.72
74
-18
4-6
-------
Table 4-6. Reported and Calculated Cement Prices in USGS Districts (2007)
USGS District
Reported
Calculated
%A
Alabama
87.09
73
-17
Illinois
91.17
72
-21
Indiana
79.99
75
-7
Kansas
92.87
71
-24
Maryland
80.18
74
-8
Missouri
89.36
71
-21
Ohio
91.47
76
-17
South Carolina
87.54
69
-21
Maine and New York
96.62
74
-24
Pennsylvania
90.85
75
-17
Michigan and Wisconsin
90.12
76
-15
Iowa, Nebraska and South
93.08
71
-23
Dakota
Florida
95.55
66
-31
Georgia, Virginia and West
93.60
74
-20
Virginia
Kentucky, Mississippi and
88.90
71
-20
Tennessee
Arkansas and Oklahoma
97.74
88
-10
Texas
88.33
80
-10
Arizona and New Mexico
86.64
101
17
Colorado and Wyoming
92.73
89
-4
Idaho, Montana, Nevada and
95.25
83
-13
Utah
California
100.24
81
-20
Oregon and Washington
90.72
75
-18
Specifically, the Arizona and New Mexico USGS district has a price differential due to
capacity constraints in New Mexico. New Mexico has only two small, veiy old kilns with
a capacity half than the demand in New Mexico. The balance demand is met by kilns in
Texas, and imports, which incur a significant transportation cost, resulting in a high
price for cement in this market. Similarly, Oregon and Washington USGS District is
dependent on supplies from other states. While demand in Washington is met by
domestic production and imports from Canada, Oregon relies on imports from Canada
and other nearby states, resulting in higher cost.
Generally, individual market prices are within the criteria specified in the QAPP
document, the deviations are explained by the demand-supply gap and transportation
cost. Moreover, root-mean-square values of the price difference for 2005 and 2006 are
about 12 percent and 10 percent respectively, and 18 percent for 2007.
Further while calibrating for price, aggregate production level is also tracked to make
sure that it is within reasonable limits. The reported and predicted aggregate production
levels for 2005-2007 are shown in Table 4-7.
4-7
-------
Table 4-7. Aggregate production (reported and modeled) for 2005-2007
Reported
Calculated
Year
Production
Production
%A
(million tons)
(million tons)
2005
111.84
112.54
0.63
2006
110.30
111.23
0.84
2007
108.17
101.95
-5.75
The current set of calibration constant values are averaged over the years of
calibration, and are available in the "Calibration" worksheet in the "ISIS_Inputs.xls"
workbook.
4.4 Recommendations
If any of the key input parameters, specifically those relating to production quantities
and costs, are refined or otherwise modified or additional observed data becomes
available, the calibration of the module should be repeated. Transportation matrix,
modes, and cost of transportation also have significant impact on the behavior of
production distribution across the districts and production prices. Therefore, if any
further modifications or refinements are made to the transportation matrix, the module
needs to be re-calibrated.
At the time of calibration of the current version of the module, production, import and
price values were used. As discussed above, when the new values become available, the
module needs to be calibrated again. Calibration of the module should be repeated as
soon as new information or new observed data becomes available. Due to practical
limitations it is recommended that the calibration of the module be repeated every two
years. Further, since the calibration data is available only for three years, equal weight
was given to the parameters obtained for each year. Once a larger data-set is available,
a modified weighing system can be adopted to give highest weight to the most-recent
year data.
4-8
-------
Appendix A
Andover Technology Partners and
RTI International Memos
Andover Technology Partners
May 7, 2010 Electrical Load for Wet Scrubbers
February 26, 2010 Wet Scrubber Cost Algorithms
July 10, 2009 GHG Mitigation Methods for Cement
Costs and Performance of Controls - revised from comments
March 10, 2009
March 10, 2009
NOx, SO2 and CO2 Emissions from Cement Kilns (Emissions
Memo) - revised from comments
RTI
March 31, 2010
August 31, 2007
ISIS Cement Production Costs
Documentation for Portland Cement Kiln Cost Functions (2005)
A-l
-------
Appendix B
ISIS Mathematical Framework
B-l
-------
Appendix B
ISIS Mathematical Framework
U-ISIS is a sector-based dynamic linear programming model that can determine optimal
sector operation for meeting demand and pollution reduction requirements over
specified time periods. The objective in U-ISIS simulation is to maximize total surplus
(see Figure B-1) over a horizon of interest for an industry, which, in general, can be a
multi-product one.
The general concept of spatial price equilibrium (SPE) in a network, where the mutual
influences of production, transportation, and consumption patterns are given fall
consideration, has been developed over the past 6 decades. In SPE network models,
interregional economies are simulated by finding the balance of demand, supply, and
trade that will result in competitive market equilibrium among the regions. Enke (1951)
first demonstrated how the cost of transportation might be included in an equilibrium
analysis of spatially separated markets by means of analogy with resistance to the flow
of current in an electric circuit. Shortly after Enke, Samuelson (1952) analyzed
interregional flows of commodities and market equilibrium using a linear programming
formulation. In this type of formulation, the equilibrium for each market of a sector is
equivalent to the quantities and prices that result while maximizing the sum of
consumer and producer surpluses for each market of the sector. This sum is referred to
as the total surplus or net social payoff of the sector; McCarl and Spreen (1980) provide
interpretation and justification. The linear programming formulation of the SPE problem
was developed by Duloy and Norton (1975).
Using Figure B-l, the definition of the suppliers' surplus corresponding to a quantity Q
of a commodity is the difference between the total revenue and the total cost of supplying
the commodity. This surplus (gross profit) is given by the area under the horizontal line
Pl-P minus the area under the inverse supply curve up to point P. Similarly, the
consumers' surplus corresponding to a quantity Q is given by the area under the inverse
demand curve up to point C minus the area under the horizontal line Cl-C. This area
is the cumulative opportunity gain of all consumers who purchase the commodity at a
price lower than the price they would have been willing to pay. It is evident from Figure
B-l that the total surplus is maximized exactly when Q is equal to the equilibrium
quantity QE. This is a very useful result, as it provides a method for computing the
equilibrium.
The total surplus concept has long been a mainstay of social welfare economics because
it takes in to account the interests of both consumers and of producers (Samuelson and
Nordhaus, 1977).
The broad modules and associated information flows utilized in the U-ISIS framework
are shown in Figure B-2. While the U-ISIS structure permits accounting for multiple
products, the description below is provided relative to one product to bring out the
important elements of the formulation and not burden the reader with many details.
Also, to make the description more readable and understandable, the input data (i.e.,
those supplied by the user, or derived from user-supplied data) are shown in bold and
the variables, whose values are determined in the optimization process, are shown in
italics. This scheme helps in organizing the numerous data elements and variables, and
B-2
-------
hopefully makes it easier for the reader to relate the data element to descriptions in the
previous chapters. Finally, the names of variables and data elements have been chosen
to be self-explanatory as far as possible.
Price
Supply curve
ci-
Consumer
surplus
Equilibrium
Producer
surplus
Pl-
D em and curve
Quantity
Figure B-l. Consumer and Producer Surplus in a Market
B.l Indexes and Mappings
Before we start the description of the mathematical structure of U-ISIS, it is helpful to
define the sets of relevant entities and mappings describing the relationships among
various entities.
Indexes (one-dimensional sets J
U-ISIS data structures (sets and parameters], variables, and equations use the following
indexes:
B-3
-------
Industry Information
Products
Production capacities and
costs
Demand
Imports parameters
Transportation costs
Noted retirements and growth
Energy Intensity
Emissions
Sulfur dioxide
Nitrogen oxides
Carbon dioxide
Fuels
Primary: coal, coke,
natural gas, coal+TDF
Substitute: tires. TDF
Production Technologies
and Emission Abatement
Approaches
Availability, applicability,
cost, and performance
Policy Parameters
Time horizon
Policy type
ISIS Engine
Economic Parameters
Discount rate
Escalation rates
Demand elasticity
Economic life
Model Outputs
Base case and controlled emissions
Production, Inter-market trades, imports
Commodity and allowance prices
Retirements and capacity growth
Controls installed
Costs
Other
Figure B-2. Broad Modules and Associated Information Floivs Utilized in the ISIS
Framework
byprod is the set of b3?products generated b}' controls;
cat is the set of catalysts used with controls;
dc s the set of demand centers across the United States;
dsteps is the set of a series of steps defined for a demand curve;
ee is the set of energy efficiency measures;
EEmeasures is the set of energy efficiency measures not including "NOEE";
eekk is the set of aliased with ee;
Emutex is the set of energy efficiency compatibility;
f is the set of fuels applicable to the sector;
i is the set of all units, including existing, replacement, expansion, and new units.
Subsets ofi described below define more-specific populations of units;
iE(i) s the set of existing units;
iii is the set of all units aliased with i;
ip(ii represents the set of projected units;
iEP(i) s the set of existing and projected units ;
irepl(i) is the set of replacement units;
iExpCapfi) represents the set of expansion units;
iNeivCap(i) represents the set of new units;
ipol(i) represents the set of units included in the policy run ;
id is the set of import districts for a sector;
-------
• isteps is a set of a series of steps defined for an import curve;
• j is the set of sector-specific pollutants of interest;
• jpol(j) represents the set of pollutants included in policy;
• jj is the set of sector-specific pollutants aliased with j;
• k is the set of sector-specific controls relevant to the emission reduction policy of
interest;
• kk s the set of sector-specific controls relevant to the emission reduction policy of
interest aliased with k;
• m is the set of raw materials specific to the sector;
• n is the set of all products for a sector;
• oi is the set of origin of imports for a sector;
• r is the set of geographic regions where units are located;
• t is the set of years in the time horizon of interest;
• tpolc(t) is the set of policy years; and
• v is the set of vintage years for added capacity.
Mappings
The mappings used in U-ISIS are:
is the mapping relating a unit i to fuels f;
• iirepl(i,irepl) is the mapping between an unit i and its replacement irepl;
• kf(k,f) is the mapping relating the control k to its fuel used f;
• ri(i,r) is the mapping relating a unit i to its geographic location r;
• poltikjee(t,i,k,j,ee) is the mapping relating availability, applicability, and the ability
to change emissions of controls k and energy efficiency measures ee to unit i,
pollutant j, and year t;
• poltik(t,i>k) is the mapping relating availability and applicability of controls k to unit
i and year t;
• poltiee(t,i>ee) is the mapping relating availability and applicability of energy
efficiency measures ee to unit i and year t;
• poltirk(t, ri(i,r),k) is the mapping relating availability and applicability of controls k
to unit i located in geographic region r and year t; and
• poltiree(t, ri(i,r),ee) is the mapping relating availability and applicability of energy
efficiency measures ee to unit i located in geographic region r and year t.
B.2 Objective Function
As mentioned above, the objective function solved in the U-ISIS model corresponds to
maximizing the total surplus (or minimizing the negative of the total surplus) for the
sector of interest over the selected time horizon. This objective function is:
B-5
-------
Minimize z — ^dis(i)-tannualprcdncost(t,i,n)
•ttranscost(t,dc,n)
• timportscost(t,id, n)
+ ^dis(t) •tcontrolcost(t,i,k)
t.i.k
•teemeasurescost(t,i,ee)
t, i, ee
+ ^ dis(t)- alprice(tj) • totalemissions(t,j)
— ^ dis(t)* J*price(t,dc,n)' ddemand(t,dc,n)
(B.2.1)
t,n
dc
where the quantities appearing in the equation above are defined for year t as follows,
dis(t) is the discount factor,
tannualprodncost(t,i,n) is the annual production cost ($) for production unit i of final
product n,
ttranscost(t,dc,n) is the cost ($) of transporting the product n to demand center dc,
timportscostft, id, n) is the cost ($) of importing foreign product n in import district id,
tcontrolcost(t,i k) is each unit's total control cost ($) using control technology k,
teemeasurescostft, i, ee) is each unit's total cost ($) using the energy efficiency
measures ee,
alprice(t,j) is the allowance price input by the user,
totalemissions(t} j) is emission of pollutant j from the sector, and
product n for all markets dc.
Note that in the objective function the term with alprice(t,j) comes into effect only if the
user provides values for alprice(t,j). If these values are specified, the model runs as
described under the "Allowance Price Inputs" in a later section.
A stepwise approximation of the demand curves (FPL-PELPS, 2003) is used in U-ISIS so
that relevant area can be computed in a linear programming scheme. This
approximation is explained below.
Stepwise Approximation of Demand. Curves in U-ISIS
For clarity, the subscripts t and dc corresponding to time and demand center are
dropped in the following explanation.
The relationship between demand and price in a market dc is expressed as:
, dc, n) ¦ ddemanc^t, dc, n) is the area under the demand-price curves for
dc
B-6
-------
P(D) = PO
D
DO
(B.2.2)
where
D is the demand for the commodity with corresponding price F[demand),
o is the elasticity of demand relative to price, and
DO and PO are the initially-specified demand quantity and price, respectively.
Figure B-3 shows a representation of the above equation and also reflects how a stepwise
approximation of the price-demand curve can be created.
Price
mm
maa
Quantity
Figure B-3. Stepwise Calculation of the Demand Curve
First the range of the demand-price curve is defined using:
Dw = DO (1-range)
(B.2.3)
Dmax =D0 (1+range)
(B. 2.4)
where
range is a user-supplied parameter with value between 0 and 1. This parameter
defines the interval Dmi,, - Dma„ within which the new equilibrium demand quantity
is expected to be found. Note range should be lar^e enough to ensure that the
solution does fall in the interval Dmin - Dma„ On the other hand a smaller value of
B-7
-------
range can increase precision of the stepwise approximation. In U-ISIS a value of 0.5
is used for range.
Next, a series of steps is defined within the interval Dmin - Dma* with the width of each
step given by:
dwidth = —P"':,x "Pmin (B. 2.5)
Number of steps
where
Number of steps is another user-supplied parameter. Increasing the value of this
parameter will increase precision of the stepwise approximation, but will increase
the model size. In U-ISIS, a value of 100 is used for Number of steps.
Now the demand quantity at the center of the slice dstep inside the interval is:
^ dwidth(2dstep-l)
D<-=D-i+ L_ V—L (B.2.6)
Using the above information, the price-demand curve is determined by:
D6
P(Ddstep) = P0
DO
(B.2.7)
Finally, the approximated area under the price-demand curve is:
Number of steps
Ypistep ¥{Ddstep) (B 2 8)
dstep s=l
-------
Substituting (B.2.8) in (B.2.1), the objective function becomes:
Minimize i = Edis(t) • tannualprcdncost(t, i)
t,l
+ £dis(t) • ttranscos(t, dcj
t,dc
+Sdis(t) • timportscost(t, id)
t,id
+2>S(t) • tcontrolcost(t, i,k)
t,i,k
+ 2>s(t) • teemeasur&cost(t, i,ee)
t,i,ee
+ ^ dis(t)- alprice(tj) • totalemissions(t,j)
E
dis(t)- ^ ElasticDemand(t, dc, dstep) • dprice(demand(t, dc,dstep)
(B.2.9)
where
demandft, dc, dstep) is the demand for the commodity at the dstep111 level of the price-
demand curve for the demand center dc in year t. The construction of the price-
demand curves has been explained above.
dprice(demand(t,dc,dstep) is the price for the commodity at the dstep111 level pf the
price-demand curve for the demand center dc in year t.
The above objective function is minimized with the constraints and related equations
described in the following sections to arrive at the relevant optimum solution.
Equation (B.2.7) is used to generate the price-demand curves for all time periods and
demand centers. However, this equation needs specifications of one point on each
demand function in each time period (i.e., [PO(t, dc), DO(t, dc)]). To determine such
points, a single run of the inelastic version of the U-ISIS model (with exogenous DO(t,
dc)) is made and then the resulting shadow prices PO(t, dc) are used in Equation (B.2.7)
to generate price-demand curves for all time periods and demand centers. These curves
are used in the last term of the objective function and in the supply constraint, Equation
(B.3.1), presented in the next section.
B.3 Supply
The demand for a commodity in a market can be satisfied by domestic production and
foreign imports as follows:
^prodnquanity(t,i, dc) + ^ importedqmntity(t, id, dc) > ^\ElasticDemand(t, dc, dstep).
i id dstep
(B.3.1)
where
-------
prodnquantity(t,i,dc) is the quantity supplied from the domestic production unit i to
demand center dc in year t, and
importedquantity(t,id,dc) is the quantity of commodity received from the origin
(country) of imports oi at the domestic import district id and supplied from that
district to demand center dc in year t.
Note that transportation of quantity from production units and import districts to
demand centers is implicit in the above equation.
The sum of all quantities in year t supplied from a kiln i to various demand centers dc
must equal the production level of unit i in that year. Then,
^prodnquantty(t,i,dc) = prodn(t,i). (B.3.2)
dc
where
prodnft, i) is the production level (tons/year) of manufacturing unit i in year t.
Note that, in general, production can be from existing units, units added at a plant (i.e.,
expansion units), newer units replacing units at a plant (i.e., replacement units), and
new kilns. Treatment of these is explained in a subsequent section.
Production of a unit is constrained by its capacity. So,
prock{t,i) < capacity(t,i) (B.3.3)
where
capacity(t, i) is the capacity (tons product/year) of manufacturing unit i in year t.
The sum of all imported quantities supplied from an import district id to various demand
centers dc in year t must equal the imports available at that import district in that year.
Then,
^ importedqmntity(t,id,dc) = ^ Imports (t,oi, id, is tep) (B.3.4)
dc oi,istep
where
Importedquantity(t, id, dc) is the imported quantity supplied from import district id to
demand center dc in year t, and
Imports(t, oi, id, istep) is the imported quantity available at the import district id in year
t at the istep111 level of the imports curve for country oi. Construction of the country
(or region) and import district specific imports-cost curves is explained below.
Stepwise Approximation of Imports Curves in ISIS
B-10
-------
The treatment of imports is similar to the treatment for elastic demand curves. Again,
for clarity, the subscripts t, oi, and id corresponding to time, origin of imports, and
import district are dropped in the following explanation.
The relationship between imports arriving from oi at id and their price is expressed as:
/
Imports
Iprice(Imports) = IpriceC
where
i
\—
ImportsO
(B.3.5)
Imports is the imports of the commodity arriving from oi at id with corresponding
imports price (value) Iprice,
a is the elasticity of imports relative to imports price, and
(ImportsO, IpriceO) is a point on the applicable quantity-price curve.
First the range of the imports-price curve is defined using:
Imin = ImportsO • (1 - range) (B.3.6)
IiiiaA = ImportsO • (1 + range) (B.3.7)
where
range is a user-supplied parameter with value between 0 and 1. This parameter
defines the interval Imin - Imax within which the new equilibrium imports quantity is
expected to be found. Note range should be large enough to ensure that the solution
does fall in the interval I min " Imax. On the other hand a smaller value of range can
increase precision of the stepwise approximation. In U-ISIS a value of 0.5 is used for
range.
Next, a series of steps is defined within the interval Imin - Imax with the width of each step
given by:
Iwidth = Im"x "Iglin (B.3.8)
Number of steps
where
Number of steps is another user-supplied parameter. Increasing the value of this
parameter will increase precision of the stepwise approximation, but will increase
the model size. In U-ISIS, a value of 100 is used for Number of steps.
Now the import quantity at the center of the slice istep inside the interval is:
isfm T Iwidth(2istep-1)
Imports = Inrfn + ^ — (B.3.9)
B-l 1
-------
Using the above information, the imports-price curve is determined by:
Imports1
Iprice(Importslstep) = IpriceO
ImportsO
(B.3.10)
Equation (B.3.10) is used to generate the imports-cost curves for all time periods, origins
of imports, and import districts. However, this equation needs specifications of one point
on each import function in each time period (i.e., [Iprice0(t, oi, id), Imports0(t, oi, id)]).
To determine such points, a single run of the inelastic version of the U-ISIS model (with
exogenous Imports0(t, oi, id) corresponding to capacities of import districts) is made
and then the resulting Iprice0(t, oi, id) are used in Equation (B.3.10) to generate the
imports-cost curves for all time periods, origins of imports, and import districts.
B.4 Production Capacity and Supply Costs
The U-ISIS model includes constraints for ensuring that endogenous production
capacity changes occur in a realistic way. This section describes how the capacity
changes take place in the U-ISIS framework and the treatment of related costs. Note
that various cost elements (e.g., capital cost of a unit [$/ton clinker] in the cement
sector) are escalated appropriately to reflect values in years of interest.
Production Capacity Related Constraints
Added capacity in year t is given by:
When t = 1,
addcap(t,i) = T^ tcap(t,i,v) • capacity(t. i)i
V
vyear(v) < tyear(t)< (vyear(v) + techlifeplants);
i e [irepl(i)^jiExpCap(i) ^jiNewCap(i)] (B.4. la)
For t > 1,
addcap(t,i) = addcap(t - timeblockj) + tcap(t,i,v) • capacity(t,i)L
V
vyear(v) < tyear(t)< (vyear(v) + techlifeplants); (B.4. lb)
i e [irepl(i)^jiExpCap(i) ^jiNewCap(i)]
where
v is the set of vintages of an unit I,
AddCap (t,i) is the added capacity of an unit i in year t,
tcap(t, i, v) is a binary variable that can bring Vth vintage of unit i online in year t,
vyear(v) and tyear(t) are parameters with values corresponding to years in the
selected time horizon,
timeblock is the block of years used in simulation, and
B-12
-------
techlifeplants is the technical life of a unit.
Only one vintage of a unit is possible for the period starting when the vintage comes on
line and ending with its technical life.
^tcap(t,i,v)< 1;
t,v
vyear(v) < tyear(t)< [vyear(v) + techlifeplants J
i e [irepl(i)^iExpCap(i) ^iNe\^ap(i)] (B.4.2)
The annual costs associated with meeting the demand for the commodity include: (1)
annualized capital costs associated with capacity growth (i.e., replacement units,
expansion units, and new capacity) and projected units, (2) annual fixed operation and
maintenance (FOM) costs, (3) annual variable costs associated with use of labor, raw
material, fuel, electricity, and operation and maintenance, (4) annual transportation
costs, and (5) annual cost of imports. These costs are described below.
Capital Recovery
If the existing units are paid for and do not have capital recovery costs, then:
plantcapcast(t,i) = 0; ie{iE(i)} (B.4.3)
where
plantcapcost(t,i) is the annual capital cost of a unit.
For projected units, for which startup date is known, annual capital cost is given by:
plantcapcost(t,i) = CRFplant • capacitv(t.i) pcapcostt(t.i):
tstart(i)< tyear(t)< (tstart(i)+ecolifeplaits);i e [iP(i)]. ^ ^
where
pcapcostt(t,i) is the capital cost (e.g., $/ton of clinker for the cement sector) of ith
unit in year t, and
CRFplant is the capital recovery factor calculated using an appropriate interest rate
and time period, ecolifeplants, for capital recovery.
Annual capital costs for all populations except existing and projected is:
plantcapccst(t,i) = AddCap(t,i) • capcosttft i)- CRFplant (B.4.5)
where
capcostt(t,i) is the capital cost (e.g., $/ton of clinker for the cement sector) for the
ith unit.
B-13
-------
Variable Costs
The annual variable cost at a unit is calculated using:
varcost(t,i) = [RMTt(t, i)+ VOMt(t, i)+ LBRt(t,i)+ ELCostt(t,i)] • prochit, i) +
fc(t, i) + H20consumptncost(t,i) + ^SWdispcost(t,i,SW). (B.4.6)
sw
where
is the raw material cost ($/ton product) at unit i in year t,
is the cost of operation and maintenance ($/ton product) at unit i in year
t,
is the cost of labor ($/ton product) at unit i in year t,
ELCostt(t,i) is the cost of electricity use ($/ton product) at unit i in year t,
fc(t, i) is the cost of fuel ($) at unit i in year t,
varcostft, i) is the annual variable cost ($) at unit i in year t,
sWdispcostftfiSW) is the annual variable cost ($) of solid waste disposal at unit i in
year t, and
H20consumptncost(t, i) is the annual variable cost ($) of water consumption at unit i
in year t.
The fuel cost for a unit is calculated as follows:
= ^ [eintensityj)- fuelcostt(t,i,f) •
f
y varprodn k, eej] - .
ni,poltikjee(t,i,k,j,ee) Nllttlber of pollutants (B.4.7)
where
eintensity(i) is the energy intensity (MMBtu/ton product) for unit i,
fuelcostt(t,i, f) is the cost of fuel f ($/MMBtu) at unit i in year t, and
varprodnft, i,f n%j, k, ee) is a production coefficient described in the next section.
Annual water consumption related cost is given by:
H20consumptncost(t,i) = H20coiisiimptii(t,i) • H20costt(1)/1000 (B.4.8)
With annual water consumption given by:
H20consumptn(t,i) = H2Ginteiisity(i)* procbi^tj). (B.4.9)
where
H20costt(t) is the cost of water ($/1000 gallon) in year t, and
H20intensity(i) is the gallons of water needed to produce a ton of product at unit i.
Annual solids discharge related cost is given by:
B-14
-------
SWdispcosl(t,i,sw) = SWgen(t,i,sw) ¦ SWdisposalcostt(t,sw) (B.4.10)
With annual solids discharges given by:
SWgen(t,i,sw) = SWgeneration(i,sw)-prochit.i). (B.4.11)
where
sWgenft, i,sw) is the annual variable of solid waste generation rate (tons/year) at unit
i in year t,
sWdisposalcostt(t,sw) is the cost of disposing of a solid discharge sw in year t, and
sWgeneration(i,sw) is the discharge of a solid sw (tons) in the process of producing
a ton of product at unit i.
Note that sWdisposalcostt(t,sw) value can be positive (disposal cost) or negative (sale
price).
Total Annual Cost of Production
Using the above information, the total annual cost of production at unit i in year t is:
tannualpralncost(t,i) = pIcmtcapcost(t,i) + varcost(t,i). (B.4.12)
Annual Cost of Imported Commodity
The cost of imports of commodity at import district id in year t is calculated using the
following equation:
timportscost(t,id) =
[(Iprice(t,oi,id,isteps)+ InsFreight(id) + imphaiidlingt(id)) Imports{t. 01. id. isteps)].
01 ,isteps
(B.4.13)
where
Iprice(t,oi,id,isteps) is the price (value) of importing commodity from origin oi
(foreign country) at the import district dc, at the istep111 level of the relevant price-
import curve,
InsFreight(id) is the insurance and freight at the import district, and
imphandlingt(id) is the handling cost at the import district.
Total Annual Transportation Cost
The cost of transporting the commodity from unit i and import district id to demand
center dc is calculated using:
B-15
-------
ttranscosft,dc) = ^[prodntransportcost1(t,i,dc) prodnquantit)(t, i, dc)] +
i
V [imprttrainportcostee) [1 -cp(i,j, k)/100]. (B.5.1)
modeintensity(i, k,ee)=
eintensity(i) + primaryHIc hange(i, k)+secondaryHIchange(i, k)+eHIdispi(i,ee) (B.5.2)
where
polfuel(t,i,f,j,k,ee) is emission (tons pollutant per ton product) of pollutant j
resulting from processing (firing) fuel f and application of any control k and/or
energy efficiency measure ee at unit i in year t, taking into account whether a unit
is available for operation in that year;
emintensityfuel(i,f,j) is the emission intensity (tons pollutant per ton product) of
pollutant j resulting from processing (firing) fuel f at unit i, taking in to account
whether any controls (e.g., Best Available Control Technology [BACT]) are already
installed on the unit;
modeintensity(i,k,ee) is the modified energy intensity (MMBtu per ton product)
needed to produce 1 ton of product at unit i. This energy intensity takes into account
any heat input changes accompanying a control k and/or an energy efficiency
measure ee;
eintensity(i) is the energy intensity (MMBtu per ton product) needed to produce 1
ton of product at unit i;
primaryHIchange(i,k) is the change in primaiy heat input (MMBtu per ton product)
due to application of control k at unit i;
B-16
-------
secondaryHIchange(i,k) is the change in heat input (MMBtu per ton product) due
to any secondary fuel addition resulting from application of control k at unit i;
eHIdispi(i,ee) is the amount of heat input (MMBtu per ton product) displaced
(reduced) due to application of energy efficiency measure ee at unit i; and
cp(i,j,k) is the reduction efficiency for pollutant j using control k at unit i.
Similarly,
polrmt(t, i, m, j, k) = e niinte us it\ rmt(i, m, j)
1 + primary RMT change perce nt(i,k)/100 +
secondary RMTchange perce nt(i, k)/100
where
[l-cp(i,j,k)/100].
(B.5.3)
polrmt(t,i,m,j,k,) is emission (tons pollutant per ton product) of pollutant j resulting
from processing raw material m and application of any control k and/or energy
efficiency measure ee at unit i in year t, taking into account whether a unit is
available for operation in that year;
emintensityrmt(i,m,j) is the emission intensity (tons pollutant per ton product) of
pollutant j resulting from processing raw material m at unit i, taking into account
whether any controls (e.g., BACT) are already installed on the unit;
primary RMTchangepercent(i,k) is the percent change in primary raw material
input (tons raw material per ton product) due to application of control k at unit i;
and
secondaryRMTchangepercent (i,k) is the percent change in raw material input
(tons raw material per ton product) due to any secondary raw material addition
resulting from application of control k at unit i.
Finally,
pol (t,i, f, m, j, k, ee) = pol fuel (t,i, f, j, k, ee)+polr mt(t,i, m, j, k). (B. 5.4)
where
pol(t,i,f,m,j,k,ee) is emission (tons pollutant per ton product) of pollutant j resulting
from processing fuel f and raw material m, and application of any control k and/or
energy efficiency measure ee, at unit i in year t, taking into account whether a unit
is available for operation in that year.
Emissions of pollutant j at unit i in year tare given by:
emissiondt, i, j) = m, j, k, ee) • varprodrit, i, f, m, j, k, ee). (B. 5.5)
X m,poltikjee(t,i,k,j,ee)
where
B-17
-------
vctrprodn(t)if)m,j>k>ee) is the production variable associated with use of kth control
and/or eeth energy efficiency measure for jth pollutant at unit i using fuel f and raw
material m in year t;
emissionsft, i,j) are the emissions (tons of pollutant per ton of clinker) of pollutant j at
unit i; and
poltikjee(t,i>k,j,ee) is the mapping described above.
Now total emissions from all units are:
totalemissons(t, j) = ^emissions(t,i, j). (B.5.6)
I
where
totalemissionsft, ,j) are total emissions (tons of pollutant per ton of clinker) of
pollutant j resulting from production activity in the entire sector.
Note that production associated with each pollutant is the same and therefore:
prodnpo(t,i, = prodnpo(tJ, f,m,jj); j * jj\ (B.5.7)
with
prodnpol(t,i,f,m,j) = ^varprodr(t,i,f,m,j,k,ee). (B.5.8)
poltikjee(t,i,k,j,ee)
where
prodnpol(t} i,f nxj,) production variable associated with pollutant j.
B.6 Controls and Costs
The U-ISIS framework includes constraints to ensure that endogenous applications of
sector-based controls and energy efficiency options occur in a realistic manner. A
description of these constraints and costs for controls is presented in this section. The
treatment of energy efficiency measures is described in a subsequent section.
Controls Related Constraints
Only one vintage of a control on a unit is possible for the period starting when the
vintage comes on line and ending with its technical life. After the technical life of the
vintage, it cannot be used.
T"\ts_c(poltik(t,i,k),v) < 1;
vyear(v)< tyear(t) < [(vyear(v)+ techlifecontrols(k)]. (B.6.1)
B-18
-------
where
v is the set of vintages of a control for unit i,
vyear(v) and tyear(t) are parameters with values corresponding to years in the
selected time horizon,
ts_c(% i, k, v) is a binary variable that can bring Vth vintage of control k for unit i online
in year t, and
techlifecontrols(k) is the technical life of control k.
Control capacity is given by the following equations.
For t = 1,
ControlCap(t,i,k) = ^ts_c(t,i,k,v)- capacity(t,i^
v (B.6.2a)
\year(v) < tyear(t) < (vyear(v) + techlifecaitrols(k)).
For t > 1,
ControlCap(t,i,k) = ControlCap(t — timeblock,i,k) + ^ ts_c(t,i,k,v) • capacity(t,i)L
v (B.6.2b)
vyear(v) < tyear(t) < (vyear(v) + techlifecaitrols(k)).
In any year, no two incompatible controls can coexist on a unit,
ControlCap(t, i, k) + ControlCap(t, i, kk) < capacity^ i) (B. 6. 2c)
where
k and kk are incompatible controls.
Finally, the production coefficient associated with a control is less than the capacity of
the unit the control is installed on,
^ varprodn(ti,f,m,j,k,ee)
-------
clinker] for the cement sector) are escalated appropriately to use values in years of
interest.
Capital Recovery and Fixed Cost
Annual recovery of capital cost of control k is given by:
capcost_c(poltik(% i,k)) =
ControlCap(t, i,k) • cntrlcapit alcostt(t,i, k) • CRFcontrol (k);
CRFcontrol(k)- ^[vyear(v)< tyear(t) < [vyear(v)+ ecolifecontrols(k)].] ^
Similarly, annual FOM cost is given by:
FOMcost_c(poltik(t, i,k)) = ControICap(t,i,k)-cntrlfixedcostt(t,i,k);
vyear(v)< tyear(t) < [vyear(v) + techlifecontrols(k)],vcntrlfixe dcostt(v, i, k)
ts_c(t, i, k, v) • y capacityjt, i)
Z
(B.6.5)
where
cntrlcapitalcostt(t,i,k) is the capital cost ($/ton of clinker) of application of control
k on ith unit,
cntrlfixedcostt(t,i,k) is the annual fixed operation and maintenance (FOM) cost
($/ton of clinker) of application of control k on ith unit,
capcost_c(t,i,k) is the annualized capital cost ($) of kth control application on ith unit,
FOMcost_c(t,i,k) is the annual fixed operation and maintenance cost ($) of kth control
application on ith unit, and
CRFcontrol(k) is the capital recovery factor calculated using an appropriate interest
rate and time period, ecolifecontrols(k), for capital recovery.
Variable Costs
Change in fuel cost associated with application of controls is given by:
fuelcostchange_c(poltirk(t,ri(i,r),k)) =
[primaryHIc hange(i, k) ^ varprodnft i,f m,j,k,ee) • fuelcostt(t, r, i)
ji(i,f),m,poltikjee(t,i,k,j,ee)
+ secondaryHIchange(i, k)- ^ varprodn,j,k,ee) • fuelcostt(t,r, f)]
kf(kJ),m,poltikjee(t,i,k,j,ee)
Number of pollutants g gj
where
primary HIc hange (i, k) is the primary heat input change (MMBtu per ton clinker)
with use of kth technology,
secondaryHIchange(i,k) is the primary heat input change (MMBtu per ton clinker)
with use of kth technology,
fuelcostt(t,r,f) is the regional cost of fuel ($/MMBtu) in year t,
B-20
-------
ri(i,r) is a set with elements corresponding to mapping between kiln and their
geographic locations,
is a set with elements corresponding to mapping between fuels and
technologies, and
poltirk(t, ri(i,r),k) is a mapping described above.
Change in raw material cost associated with application of controls is given by:
RMTcostchange_c(poltirk(t, ri(i, r),k)) =
[primaryRMT changepercent /100 • ^ varprodn(t,i,fm,j,k,ee) ¦ RMTt(t,i)
Ji(i,f),m,poltikj2e(l,i,k,j,ee)
+ secondaryRMTchange percent /100 • ^ varprodn(t,i,fm,j,k,ee) • RMTt(t,i)]
Ji(i.f ).m.poltikjee(t.i.k.j.ee)
Number of pollutants ^ yj
where
primaryRMTchangepercent is the percent change in primary raw material with use
of kth technology,
secondaryRMTchange(i,k) is the percent change raw material input corresponding
to any secondary raw material addition with use of kth technology, and
is the unit-specific cost of raw material ($/ton clinker) in year t.
Annual reagent consumption costs are:
rgntconsurrptcost _c{poltik(t,i,k)) =
^ rgntconsurpt _c(t, i, k, rgnf) • reagentpricet(t,rgnt).
rgnt (B.6.8)
The annual consumption of a reagent given by:
rgntconsumpt _ c(poltik(t, i, k), rgnt) =
y [reagentconsumptfuel(i, f, k,j, rgnt) • ^yarprodn(ti,fm,j,k;ee)]
m,poltikjee(tJ,k,j\ee)
Number of pollutants
+
y [re agentcons umptrmt(i ,m, k,J, i"gnt) • ^ varprodn(t i,fm,j,kee)]
m,j fi(i,f),poltikjee(tJ,k,j\ee)
Number of pollutants g gj
where
reagentconsumptfuel(i,f,k,j,rgnt) is the reagent consumption due to control (tons
reagent/tons clinker) associated with fuel-based emission intensity,
B-21
-------
reagentconsumptrmt(i,m,k,j,rgnt) is the reagent consumption due to control (tons
reagent/tons clinker) associated with raw material-based emission intensity,
reagentpricet(t,rgnt) is the price of reagent rgnt ($/ton of reagent) in year t, and
rgntconsumpt_c(t} i, k, rgnt) is the annual reagent consumption due to control (tons
reagent/year), and
Rgntconsumptcos^cfoik) is the annual reagent consumption cost ($/year) due to
control k at unit i in year t.
Catalyst consumption cost is:
catconsumptcost _ c{poltik(t, i, k)) =
y\catconsumpt _ c(t, i, k, cat) • catalystpricet(t£at);
iZ (B.6.10)
with
catconsumpt _ c(poItik(t, i, k),cat) =
^catalystconsumptO, k,j,cat) EGFW(j) . ^varprodnM>»J.kee)
j 10000 Jj(i,f),m,poltikjee(t,i,k,j,ee)
Number of pollutants (B 6 11)
where
catalystconsumpt(i,k,j,cat) is the catalyst consumption rate (ft3 catalyst/10000 ft3
flue gas),
EGFW(i) is the exhaust gas flow rate (scf/ton clinker),
catalystpricet(t,cat) is the price of the catalyst ($/ft3) in year t,
catconsumpt_c(t, i, k, cat) is the catalyst consumption rate (ft3/year), and
catconsumptcost_c(t,i,k) is the catalyst cost ($/year).
Annual cash flow associated with byproduct generation disposal/sale is:
byproductcost _c(poltik(t,i,k)) =
^ byproductgsn _ c(t,i,k, by prod)-byproductpricet(t.byprod);
(B.6.12)
with annual generation of a byproduct given by:
B-22
-------
byproductgsn _ c(poltik(t, i, k), byprod) =
/byproductp rodnfuel(i ,f, k,j, byprod) • ^ varprodn(ti,f m,j,k,ee)]
m ,poltikjee(t j ,k, j ,ee)
Number of pollutants
+
^ /by productp rodnrmt(i, in, k,J, byprod) • ^ varprodn(ti,f m,j,k,ee)]
mj Ji(i,f),poltikjee(t,i,k,j,ee)
Number of pollutants (B 6 13)
where
byproductprodnfuel(i,f,k,j,byprod) is the byproduct generation due to control (tons
reagent/tons clinker) associated with fuel-based emission intensity,
byproductprodnrmt(i,m,k,j,byprod) is the byproduct generation due to control
(tons reagent/tons clinker) associated with raw material-based emission intensity,
byproductpricet(t, byprod) is the price of disposing or selling the byproduct byprod
($/ton of reagent) in year t,
byproductgenjcfci,k,byprod) is the annual byproduct generation due to control (tons
reagent/year), and
byproductcost_c(t} i, k) is the annual cash flow associated with byproduct
generation/sale due to control k at unit i in year t.
Note that byproductpricet value can be positive (disposal cost) or negative (sale price).
Annual electricity consumption (kWh/yr) due to control is:
^ [kWhperton(i,f,k) • y^yarprodn(ti,fm,j,k,ee)\
kwh_C(poltlk{t, i9 k)) = — ^polt^it^ee) _
Number of pollutants ^ ^
The cost of electricity consumption is:
kwhcost_c(poltik(tJ,k)) = kwh_c(t,i,k) • electricitxrostt(t.i). (B.6.15)
where
kwh_c(t, i, k) is the annual electricity consumption (kWh/year) due to control k at unit
i in year t,
kwhcost_c(t, i, k) is the annual electricity consumption cost ($) due to control k at unit
i in year t,
KWhperton(i,f,k) is the electrical requirement (kWh per ton of clinker) for technology
k, and
electricitycostt(t,r) is the electricity price ($/kWh) in year t at unit i.
Annual water consumption associated with control k is:
B-23
-------
HlOconsumpt_ c(poltik(% i,k))
y /H20usefuel(i, f, k) £ varprodn(t,i,f,m,j,kee)]
m,poltikjee{t j ,k,j ,ee)
Number of pollutants
+
y /H20usermt(i, m, k) ¦ ^ varprodn(t,i,fm,j,k,ee)]
m,j fi(i,f),poltikjee(t,i,k,j,ee)
Number of pollutants
The cost of water consumption is:
H20cost_c(poltik(t,i,k)) = H20consumpt_c(t, i,k) - H20costt(t) /1000
where
H20usefuel(i,f,k) is the water use due to control (tons reagent/tons clinker)
associated with fuel-based emission intensity,
H20usermt(i,m,k) is the water use due to control (tons reagent/tons clinker)
associated with raw material-based emission intensity,
H20costt(t) is the price of water ($/1000 gallons) in year t,
H20consumpt_c(t)ik) is the annual water use due to control (gallons/year), and
H20cost_cft>i,k) is the annual cost of water consumption ($) due to control at unit i
in year t.
Total Annual Cost
Using the above costs, the total annual cost of controls is:
tcontrolcost(t,i,k) = capcost_c(t,i,k) + FOMcost_c(t,i,k)
+ fuelcostchange_c(t,i,k) + RMTcostchange_c(t,i,k)
+ rgntconsur?ptcost_c(t,i,k) + catconsumptcost_c(t,i,k)+ byproductcost_c(t,i,k)
+ kWhcost_c(t,i,k) + H20cost_c(t,i,k) (B.6.18)
B.7 Costs of Energy Efficiency Measures
As described below, the treatment of energy efficiency measures in U-ISIS is similar to
that for control.
Energy Efficiency Measures Related Constraints
As for controls, constraints are needed to ensure realistic applications of energy
efficiency measures. These constraints are described below.
(B.6.16)
(B.6.17)
B-24
-------
Only one vintage of an energy efficiency measure on a unit is possible for the period
starting when the vintage comes on line and ending with its technical life.
^ts_ee(poltiedjt,i,ee),v) < 1;
t,v
vyear(v) < tvear(t)< [(vyear(v) + etechlife(ee)]. (B.7.1)
where
v is the set of vintages of measure ee,
poltiee(tpolc(t) is the mapping described before,
vyear(v) and tyear(t) are parameters with values corresponding to years in the
selected time horizon,
ts_ee(t, i, ee, v) is a binary variable that can bring Vth vintage of measure ee online on
unit i in year t, and
etechlife (ee) is the technical life of measure ee.
Energy efficiency measure capacity is given by the following equations.
For t = 1,
EECap(t, i, ee) = Z>- ee(poltiedt,i,ee),v)- capaci ty (t,i\
V
vyear(v) < tyear(t) < (vyear(v) + eetechlifijee)).
For t > 1,
EECap(t,i,ee) = EECap(t — timeblock,i,ee) + 2>- ee(poltiee(t,i,ee),v) • capacity(t,i^
V
vyear(v) < tyear(t)< (vyear(v) + eeteclilif^ee)).
In any year, no two incompatible measures can coexist on a unit,
Ik ee(poltiee(t, /, ee), v) + ts _ee(poltiee(t, i, eee),vj]< 1; (B.7.2c)
V
where
ee and eee are incompatible measures.
Finally, production coefficient associated with an energy efficiency measure is less than
the capacity of the unit the measure is installed on,
7, varprodn ee)
f,m,poltilgee(t,i ,k,j, ee)
(B.7.2a)
(B.7.2b)
-------
capacity(t,i) is the capacity of unit i in year t.
Capital Recovery and Fixed Cost
For an energy efficiency measure ee, annual recovery of capital and annual FOM cost
are given by Equations B.7.4 and B.7.5, respectively.
capcost_ee(poltiee(t, i,ee)) =
EECap(t,i,ee)¦ ecapitalccstt(t,i,ee)- CRFEE(ee);
vyear(v) < tyear(t) < [vyear(v) + ecolifeee^e)]. (B.7.4)
FOMcost _ee(poltiee(t, i,eej) =
EECap(t,i, ee)-efixedcostt(t,i»ee),
vyear(v) ee) is the capital cost ($/ton of clinker) of eeth energy efficiency
measure application on ith unit,
efixedcostt(t,i>ee) is the annual fixed operation and maintenance (FOM) cost ($/ton
of clinker) of eeth energy efficiency measure application on ith unit,
ts_ee(t,i,ee,v) is a binary variable that can bring Vth vintage of measure ee online on
unit i in year t,
capcost_ee(t,i,ee) is the annualized capital cost ($) of ee application on ith unit, and
FOMcost_ee(t, i, ee) is the annual fixed operation and maintenance cost ($) of ee
application on ith unit, and
CRFEE(ee) is the capital recovery factor calculated using an appropriate interest
rate and time period, ecolifeee(ee), for capital recovery.
Variable Costs
Change in fuel cost associated with application of ee measures is given by:
fuelcostchange _ ee(poltiree(t, ri(i,r),ee)) =
eHIdispi(i, ee) • ^ varprodn (ii,fmj, k, ee) - fuelcostt(t, r, f)
f,m,poltilgee (t,i Jc,j,ee)
Number of pollutants 7 gj
where
eHIdispi(i,ee) is the displacement of primary heat input (MMBtu per ton clinker)
with use of eeth energy efficiency measure,
fuelcostt(t,r,f) is the regional cost of fuel ($/MMBtu) in year t,
ri(i,r) is a set with elements corresponding to mapping between kiln and their
geographic locations, and
poltiree(t, ri(i,r),ee) is a mapping described above.
Annual electricity consumption (kWh/yr) due to an energy efficiency measure is:
B-26
-------
ekWhperton(i,ee) • ^\arprodn(ti,f,mj,k,ee)
kwh _ ee(polhee{t, i, ee); =
Number of pollutants y yj
The cost of electricity consumption is:
kwhcost_ee(poltiee(t,i,ee)) = kwh_ee(t,i,ee) • electriciJ^costt(t,i) (B.7.8)
where
eKWhperton(i,f,ee) is the electrical requirement (kWh per ton of clinker) for ee, and
electricitycostt(t,i) is the electricity price ($/kWh) in year t at unit i.
Total Annual Cost
Using the above costs, the total annual cost of energy efficiency measures is:
teemeasurescost(t, i,ee) =
cap cost _ee(t,i,ee)+FOMcost _ee(t,i,ee)+
fuelcostchange _ee(t,i,ee)+kWhcost _ee(t,i,ee) (B.7.9)
B.8 Policy Options
U-ISIS can help design and evaluate a number of emissions reduction policy options
including cap-and-trade, emissions taxes, and emissions limits. Additionally,
appropriate combinations of these options can also be evaluated. The policy options
included in U-ISIS are described below.
Cap-and-Trade
Under this option, an emissions cap is set on the amount of a pollutant that can be
emitted. Sources, companies, or other groups are issued emission permits (allowances)
which represent the right to emit a specific amount of the pollutant. Allowances may be
banked for use in the future. The total amount of allowances available in the current
period and those banked in previous periods cannot exceed the cap in the current
period. Sources or companies that need to increase their emissions must buy allowances
from those who pollute less. The transfer of allowances is referred to as a trade. In effect,
the buyer is paying a charge for polluting, while the seller is being rewarded for having
reduced emissions by more than was needed. Thus, in theory, those that can easily
reduce emissions most cheaply will do so, achieving the pollution reduction at the lowest
possible cost to society.
Generally, annual caps have been utilized in ongoing programs (e.g., Title IV SO2
reduction program). However, U-ISIS does permit evaluation of potential programs with
caps over user-selected time periods (e.g., 5-yearly caps). This evaluation is
accomplished using,
B-27
-------
totalemissons(t,j) < ecap(t, j) + bnk(t, j) - bnk(t + period, j)
(B.8.1)
where
ecap(t,j) is the emission cap for pollutant j in year t, and
bnk(t+period, j) are the allowances of the pollutant j banked in year t for the year t +
period.
Cap-and-Trade with a Minimum Reduction Requirement
While designing a cap-and-trade program, there may be an interest in requiring a
minimum level of emission reduction from each affected entity. Such a requirement may
be able to help address any local emissions-related concerns. In U-ISIS, this
requirement can be imposed using:
eirissions(t,i,j) < ecpiriner(t,i,j) (B.8.2)
where
ecpminer(t,ij) is the unit-specific minimum emission reduction requirement for
pollutant j in year t.
Emissions Limits
U-ISIS permits evaluation of the costs and emissions reductions associated with more
traditional emission reduction programs utilizing unit-specific rate-based emission
limits. Such requirements are imposed using
emissions(t,i,j) < el(t4,j) • prodn(t,i) (B.8.3)
where
el(t,ij) is the rate-based emission limit for pollutant j in year t and every affected
unit i complies with this limit-
Allowance Price Inputs
In some cases, there may be an interest in endogenously determining the level of
emission reduction corresponding to a certain allowance price. This information may be
useful, for example, in a situation where an allowance price is set for reducing emissions
from many industrial sectors. In such a case the levels of emission reductions
corresponding to the same allowance price may be different for the sectors under
consideration. The emissions response to a given allowance price is driven by the
following term in the objective function (see section B. 1),
^dis(t)alprice(t j) -totalemissons(t, j) (B.8.4)
where
B-28
-------
alprice(t,j) is the exogenous allowance price of pollutant j in year t.
Note that the allowance price inputs scheme above is equivalent to emission-tax-based
programs in which affected units or companies pay a tax for eveiy unit of pollution they
produce. Thus this scheme can also be used to evaluate such programs.
B.9 Optimization and Post-Processing
In U-ISIS, the input data are pre-processed to arrive at suitable input parameters for
use in the model equations explained in this chapter. Once the data have been pre-
processed, U-ISIS solves for the appropriate levels of production, imports and controls
required to meet the constraints associated with commodity demand and emissions,
while maximizing total surplus. Once the surplus maximization problem has been
solved, the results are post-processed to obtain parameters and level values of the
variables of interest. The key variables of interest are: production level of each
production unit to meet regional demand, level of imports in each region, installation of
various controls, emissions, and various costs. Output data are written in appropriate
worksheets in an Excel workbook and further linked to various plots to enable visual
presentation and analyses of the results.
B-29
-------
References for Appendix B
Duloy, J. H.; Norton, R.D. (1975). Prices and Incomes in Linear Programming Models.
American Journal of Agricultural Economics. 57(4): 591-600.
Enke, S. (1951). Equilibrium among Spatially Separated Markets: Solution by Electric
Analogue. Econometrica. 19: 40-47.
FPL-PELPS, A Price Endogenous Linear Programming System for Economic Modeling,
Supplement to PELPS III, Version 1.1, USDA Research Paper FPL-RP-614, 2003.
McCarl, B.A.; Spreen, T.H. (1980). Price Endogenous Mathematical Programming as a
Tool for Sector Analysis. American Journal of Agricultural Economics. 62: 87-102.
Samuelson, P.A. (1952). Spatial Price Equilibrium and Linear Programming. American
Economic Review. 42(3): 283-303.
Samuelson, P.A.; and W. Nordhaus. (1977). Economics (17th edition), John Wiley.
B-30
------- |