Review of ISIS Documentation Package

Prepared for:

U.S. Environmental Protection Agency

Air Pollution Prevention and Control Division

Research Triangle Park, NC 27711

Prepared by:

ARCADIS U.S., Inc.

4915 Prospectus Drive, Suite F

Durham, North Carolina 27713

Contract No.: EP-C-04-023

Project No.: RN990234.0052

April 15, 2009



Table of Contents

  TOC \o "1-4" \u  Overview	  PAGEREF _Toc228074699 \h  3 

Reponses to Comments on Chapter 1: Introduction	  PAGEREF _Toc228074700
\h  4 

Reponses to Comments on Chapter 2: Data Used in ISIS	  PAGEREF
_Toc228074701 \h  12 

Reponses to Comments on Chapter 3: ISIS Mathematical Framework	  PAGEREF
_Toc228074702 \h  22 

Reponses to Comments on Chapter 4: Calibration of Cement ISIS Model	 
PAGEREF _Toc228074703 \h  29 

Reponses to Comments on Chapter 5: Illustrative Analysis	  PAGEREF
_Toc228074704 \h  32 

Reponses to Comments on Chapter 6: Summary	  PAGEREF _Toc228074705 \h 
35 

Reponses to Comments on Appendix A	  PAGEREF _Toc228074706 \h  36 

Reponses to Comments on Appendix B	  PAGEREF _Toc228074707 \h  45 

Reponses to Editorial Comments	  PAGEREF _Toc228074708 \h  47 

Attachment A Invitation Letter to Reviewers	  PAGEREF _Toc228074709 \h 
51 

Attachment B Review Comments from Dallas Burtraw, Resources for the
Future	  PAGEREF _Toc228074710 \h  54 

Attachment C Review Comments from Kevin Culligan, US Environmental
Protection Agency	  PAGEREF _Toc228074711 \h  60 

Attachment D Review Comments from Fereidun Feizollahi, CARB Economic
Studies Section	  PAGEREF _Toc228074712 \h  62 

Attachment E Review Comments from Sikander Khan, US Department of Energy
  PAGEREF _Toc228074713 \h  63 

Attachment F Review Comments from Andy O'Hare, Portland Cement
Association	  PAGEREF _Toc228074714 \h  69 

Attachment G Review Comments from Hendrik van Oss, US Geological Survey	
 PAGEREF _Toc228074715 \h  70 

Attachment H Review Comments from Hector Ybañez, Holcim	  PAGEREF
_Toc228074716 \h  89 

 

Overview

The ISIS Documentation Package was sent by ARCADIS via email to the
following six reviewers on December 23, 2008:

Dallas Burtraw							Kevin Culligan

Resources for the Future					US Environmental Protection Agency

  HYPERLINK "mailto:burtraw@rff.org"  burtraw@rff.org 						  HYPERLINK
"mailto:culligan.kevin@epamail.epa.gov"  culligan.kevin@epamail.epa.gov 

Sikander Khan							Andy O'Hare

US Department of Energy					Portland Cement Association

  HYPERLINK "mailto:sikander.khan@hq.doe.gov"  sikander.khan@hq.doe.gov 
				  HYPERLINK "mailto:aohare@cement.org"  aohare@cement.org 

Hendrik van Oss						Hector Ybañez

US Geological Survey						Holcim

  HYPERLINK "mailto:hvanoss@usgs.gov"  hvanoss@usgs.gov 						 
HYPERLINK "mailto:Hector.Ybanez@holcim.com"  Hector.Ybanez@holcim.com 

And to one more reviewer on January 13, 2009:

Fereidun Feizollahi

CARB Economic Studies Section

  HYPERLINK "mailto:ffeizoll@arb.ca.gov"  ffeizoll@arb.ca.gov 

A copy of the invitation letter is shown in Appendix A. In addition to
the invitation letter, the ISIS review package included the following
files:

Review documentation file – “ISIS Peer Review Documentation
12-23-2008.pdf”

GAMS and ISIS instructions – “GAMS Installation and ISIS Running
Instructions.pdf”

GAMS license file to run the trial version – “GAMSLICE.TXT”

GAMS model files and Input/Output files – “ISIS_PeerReview.zip”

The remainder of this document consists of responses to reviewers’
comments. Responses are arranged by chapter number in the “ISIS Peer
Review Documentation” and address sequentially numbered comments with
reviewer’s initials and the number of the chapter to which the comment
pertains. For example, comment number DB1.2 denotes a second comment
pertaining to Chapter 1 in the “ISIS Peer Review Documentation” and
received from Dallas Burtraw. Sequentially numbered editorial comments
from all reviewers are given in the Response to Editorial Comments
section of this document. For example, HOED1 denotes first editorial
comment from Hendrik van Oss. Complete, as-received, review comments are
given in Appendices B through H.  After each reviewer’s comment, the
reader will find EPA’s response as of the date of the Portland cement
NESHAP proposal.  EPA is currently working on a post peer-review version
of the ISIS model and is updating the model documentation to address
peer review comments.

Reponses to Comments on Chapter 1:

Introduction

DB1.1: A potentially valuable improvement in the model capability would
be to allow for emissions trading programs (or regulatory standards)
that have a location and/or time dimension…They (episodic controls)
could be implemented within an incentive-based (trading or tax) approach
to policy…Could the model be used to simulate this type of regulatory
issue by incorporating information about the cost of adjusting
production schedules or fuels? Also, could the trading program
represented in the model have a regional dimension…?…Can this
(episodic operation of the facility) be addressed in the narrative now
and in a future version of the model? 

Not relevant for NESHAP, but capability to take into account episodic
trading, controls and operation will be added to the next version of the
model.

DB1.2: An important dimension of environmental impact of a facility is
its stack characteristics. These are not mentioned anywhere. What is the
stack height? 

To be addressed in the updated ISIS model/documentation.

DB1.3: Can the model accommodate a bank that does not expire before the
end of the second period, or is a terminal year date necessary for the
banking algorithm to converge? How would the terminal year be
determined? Is there a backstop technology that determines long-run
abatement cost?

Model related—current version of the model uses a banking algorithm
based on standard bank-exhaust condition at the end of the policy
horizon. However, we will consider incorporation of appropriate backstop
technology to establish long-run abatement cost or steady-state
equilibrium condition for the bank in the long-run. 

Same as SK5.2

DB1.4:…the threat of unfair international competition under climate
policy…Since this is an emerging issue it would be a great enhancement
to have it considered and modeled carefully in ISIS.

Model of Imports- - we will review modeling of imports and consider
possible enhancements in the model to incorporate impacts of
international competition.

DB1.5:…NOx “trimming”…The background technical memos discuss it
somewhat, but it does not appear in the model documentation.

NOx “trimming” meaning that small changes in combustion temperature
are used to reduce the formation of heat related NOx. For example, by
reducing excess air from 10 to 5% (oxygen exhaust levels from 2 to 1%),
NOx emissions can be reduced by approximately 15% (1994 ACT). Reducing
excess air levels results in indirect reduction of NOx emissions per
amount of clinker produced (ACT 2008).

DB1.6: Page 1-11, Figure 1-7.…emissions inputs are describing baseline
or uncontrolled emissions at baseline levels of production. This should
be explained in the context of Figure 1-7. 

Same as HO1.13

Figure 1-7 has been modified to reflect the comment

DB1.7: Are investment decisions using more than an annualized capital
cost to consider in-place controls? 

Investments decisions use annualized capital cost and fixed and variable
O&M costs.

DB1.8: It is clear from the optimization problem that emissions policy
compliance costs (taxes, permits) are included in the variable cost
calculation…It should be mentioned in Chapter 1.

To be addressed in the updated ISIS model/documentation.

DB1.9: Some short (CEMStar) summary should be included in the ISIS model
documentation.

Brief text on CEMStar and a reference were added.

DB1.10: What is the plan for updating the data inputs and technology
characteristics?…For example, Page 1-7, top…

Data inputs and technology characteristics will be updated as necessary.
For example, current expansion projection data do not account for the
economic situation of late 2008/early 2009. 

DB1.11: Page 1-5, bottom. How does one interpret “kiln capacity?” Is
output measured per unit time? I learn in the memo from Andover
Technology Partners (September 23, 2008) in a footnote on Page 1 that
capacity is normally expressed “per hour.” This should be added
throughout the ISIS documentation.

Kiln capacity comes from the PCA; U.S. and Canadian Portland Cement
Industry: Plant Information Summary. December 31, 2004

Clinker capacity definitions:

annual capacity – is daily capacity multiplied by 365 less normal
downtime days (given in metric tons of clinker)

daily capacity- is the normal clinker output a kiln can produce per day
given a realistic work pattern 

Normal downtime days - is the number of days of downtime required for
maintenance, repair, or clean up.

DB1.12: Also, the “Production Technologies…” node could have more
information in the box. Is technology choice a decision made in this
node?

Figure 1-7 has been modified. Production, Technologies and Emission
Abatem2ent Options addresses applicability, cost algorithms, and
performance. These data are inputs in ISIS; the model determines which
option is needed to meet the emission reduction requirement.

DB1.13: Page 1-9, bottom. Local constraints such as nonattainment areas
and PSD areas are described, but there is no indication whether these
are represented in the model or have been ignored.

These local constraints such as NAA areas, class I Areas are not
represented in ISIS at this time. Future version of ISIS will
incorporate these air quality considerations or constraints.

HO1.1: Section 1.1. Introduce this with portland and blended cement
(both are used in concrete), and then state that you will focus on
portland. But be careful: the USGS monthly data (1/98 onwards) split out
the two (Table 2a vs. 2b) and they should be added together for
discussions of consumption. I do not know if the PCA, in using our data,
has combined these in all cases. The USGS annual data includes blended
cement within the portland umbrella. The annual report (Mineral
Yearbook) shows the combined monthly sales (consumption) data in Table
9.

The model uses consumption data from PCA not from USGS.  Ed Sullivan,
chief economist of PCA told EPA (oral communication) that their
consumption data is for portland cement only, it doesn’t include
blended cement.

Text in Section 1.1.1 has been revised. Mineral Yearbook data for cement
production are for portland cement (including blended cement) and
masonry cement (for example 99,319,000 metric tons for 2005  shown in
Table 1).

HO1.2:…it is NOT acceptable to use “dry” to mean (only) a dry kiln
lacking a preheater and/or preheater/precalciner. You should call this
type of kiln a “long dry” kiln. The 3 general categories for dry
kilns are: 1) long dry kiln; 2) preheater kiln; 3) preheater/precalciner
kiln (which, after spelling out the first time, you may shorten to
precalciner kiln). Para 1 below Figure 1-3: Here’s where you are
tripping over your misuse of the term “Dry” kiln. First, our Table 5
did show 16.3% from wet plants, but this is NOT quite the same thing as
saying that wet kilns did 16.3% of the clinker. This is because there is
a separate line for the relatively small output from “combination”
plants (those that operate both wet and dry kilns). Likewise, the 11.8
Mt of clinker form Dry plants understates the output of the dry kilns
because a couple of these are in the line for combination plants. And
you misstate the 11.8 Mt “Dry” as being from dry kilns lacking a
preheater or precalciner. 

Information in the text has been revised to state that 2005 clinker
production was as follows: 13.5% from wet kilns, 81% from dry kilns (3
types) and 5.5% from plant operating wet and dry kilns. Data base on
USGS Mineral Yearbook 2005, Table 7.

HO1.3: It may be useful to discuss what is meant by “capacity”
anyway! Daily capacity means per 24 hrs (and implies 24 hrs of
continuous operations) and is more or less unambiguous…except that you
don’t know whether the cited number is the manufacturer’s rating on
the kiln (usually a conservative figure), or actual “full blast—but
normal operating—conditions. Commonly, the rated capacity is lower
than the actual operational capacity. Annual capacity is not well
defined. The USGS calculates an “apparent annual capacity” as
follows:

Apparent annual capacity = (Reported daily capacity) * (X-Y) where:
capacity is in tons; X is 365 days (or 366 in leap years) and Y is the
number of days of downtime for routine, scheduled, maintenance. Thus,
(X-Y) is the scheduled operational year. The PCA, I believe, uses the
same formula, but (in my opinion) does not always check to distinguish
routine maintenance downtime from total downtime. Clearly, there can be
a large variation in apparent annual capacity if there is any
mischaracterization of the routine maintenance downtime or the daily
capacity. The USGS calculates—for the plant overall—the capacity
utilization wrt. the actual clinker output: we query (question) any
utilization percentages in excess of 100%.

Because there is commonly downtime in excess of routine maintenance
(extended upgrades, shutdowns for unplanned repairs…), the USGS feels
that a utilization rate of 85% or more represents full practicable
capacity operations.

[The industry reports to the USGS, for each kiln, the daily capacity,
the total downtime, the subset downtime for routine maintenance, and the
remainder for “other downtime”. We also ask for the dimensions of
the kiln, whether it is wet or dry, whether it has a preheater, whether
it has a precalciner, and whether dust control is via a baghouse or an
ESP.]

Refer to DB1.11 – annual capacity definition is from PCA- and was
added in Chapter 2 section 2.2.1.1

HO1.4: Table 1-1: The data are typical averages, but are not universal.
The reader may think these are fixed data. The EPA source sited likely
used either USGS or PCA data.

Table 1-1 title changed to: Typical Average Heat Input by Cement Kiln
Type

Commenter is correct, data in Table 1-1 are typical averages. Data in
Table 1-1 comes from Table 3-3 in the 2007 ACT.  In turn, Table 3-3  in
the ACT quotes EPA’s 2000 document below:

U.S. Environmental Protection Agency. NOx Control Technologies for the
Cement

Industry. Prepared by EC/R Incorporated. September 19, 2000. Available
at

www.epa.gov/ttn/naaqs/ozone/ozonetech/cement_nox_update_09152000.pdf.

  HYPERLINK "http://epa.gov/ttn/catc/dir1/cement_updt_1107.pdf" 
http://epa.gov/ttn/catc/dir1/cement_updt_1107.pdf 

  HYPERLINK
"http://www.epa.gov/ttn/naaqs/ozone/ozonetech/cement_nox_update_09152000
.pdf" 
http://www.epa.gov/ttn/naaqs/ozone/ozonetech/cement_nox_update_09152000.
pdf 

HO1.5: I would note that the USGS Minerals Yearbook chapters on cement
(MYB) also allow the average capacity to be discerned: it would
calculate as 0.577 Mt/yr (million metric tons per year) in 2005 (MYB,
Table 5).

Reviewer is correct, average kiln capacity could be inferred from data
presented in Table 5 of the USGS Mineral Yearbook 2005.

HO1.6: Page 1-6: Figure 1-3: the color scheme would be improved by using
blue for wet and red(dish) for dry. Cite the data source.

Thank you. We will incorporate the suggested color change in the plot
and add the data source. 

HO1.7: Figure 1-4: this plant location map is incomplete because it
fails to adequately discern localities with more than one plant in a
cluster. For example: WA has 3 plants (2 integrated plants @ Seattle and
a grinding plant at Bellingham); CA has 11 plants; FL has 6 plants,
etc…

Map shows just integrated facilities.

Map will be updated to adequately discern localities with more than one
plant in a cluster

HO1.8: Page 1-7: your discussion of imports is misleading: imports can
be by domestic producers seeking to feed markets where domestic
(particularly their own domestic) production is inadequate (or where
they wish to capture market share) AND by independent importers who
compete against domestic production in some markets, and help feed
markets not easily serviced by domestic production.

Text was modified to address the issue of domestic producers wishing to
capture market share.

HO1.9: Figure 1-5 needs a citation for the data source (USGS I presume).
The curve for import share is misleading and represents a problem with
the USGS data: we compare the level of cement imports to that of total
consumption (apparent consumption calculated as cement production +
cement imports – cement exports – change in cement stocks). We
exclude the imports of clinker because the production of cement includes
cement made in the USA from imported clinker. But we do not include
changes in clinker stockpiles (but should) mainly because we did not
collect such data prior to a few years ago. Done “properly” the
import reliance would be 1-2% higher than what we show currently in the
data series.  

Figure 1-5 has been modified in that the “Import Share” curve has
been removed.  Data points for 2004 and 2005 agree with the ones on USGS
Mineral Yearbook but there is no older data in the Yearbook.

HO1.10: Another related issue is the cement stockpiles: the comparison
is between successive yearend (12/31) stocks. But you could have a
significant drawdown or buildup due entirely to the early arrival or
late arrival of a few ships—this would cause a stockpile shift
entirely UNRELATED to any economic forces. So I do not find apparent
consumption to be a very useful statistic.

To be addressed in the updated ISIS model/documentation.

HO1.11: Oddly enough, there is a lot of flexibility in how the industry
views the cost of imports vs. cost of local production. A company might
continue to import—even at a loss or minimal profit—for at least a
while if this is the only way they can keep their customers happy (i.e.,
not going to another supplier)…

…the data (in Table 1-3) are for imports of “hydraulic cement and
clinker”—meaning, all types of hydraulic cement, and all types of
hydraulic clinker--so they are not just portland cement imports. This
(cement and clinker) issue applies, I think, to all import data that you
use in this report and represents a fairly serious problem for your
model.

Re-label the table to imports of hydraulic cement and clinker

To be addressed in the updated ISIS model/documentation.

HO1.12: If you are going to quote the total weight of PM10 emissions,
put it in the context of something—such as the weight of the total raw
materials and fuels.

Quarrying operations (the crushing and grinding of raw materials and
clinker), along with kiln line operations result in particulate matter
(PM) emissions. The NEI estimates show that, in 2002, the sector
released 36,000 tons of PM10 emissions.

HO1.13: The text in this area ought to mention CO2. In fact, the whole
section is awfully brief—but perhaps you could refer the reader to the
discussion in the Appendix.

Discussion of CO2 emissions from cement kilns has been added ending with
the referral to Appendix A for a more detailed discussion.

HO1.14: Figure 1-7: should you add (flowsheet box for emissions) HAP and
PM? 

Same as DB1.6- 

HY1.1: Page 1-5. Statement about PC fuel efficiency in the first
paragraph - replacement of both wet and certain dry process kiln
capacity with modern kiln processes yield substantial reductions in fuel
use. This is true in term of fuel efficiency but not in the absolute
amount of fuel use. Newer kilns tend to be bigger kilns and are going to
use more fuel.

Done

HY1.2: Page 1-7. Figure 1-4 Add import’s custom districts to the map

Map will be updated to adequately discern localities with more than one
plant in a cluster and add custom districts

HY1.3: Page 1-8. Do these imports include imports from Mexico (imports
by rail and trucks)?

Data in Table 1-3 on page 108 do not include imports from Mexico, other
than for Miami. This is consistent with Table 18 in the USGS Mineral
Yearbook (2005).

HY1.4: Page 1-9. Are the PM emissions (36,000 tons) emissions from the
kiln or for the whole plant (including grinding)

Quarrying operations (the crushing and grinding of raw materials and
clinker), along with kiln line operations result in particulate matter
(PM) emissions. The NEI estimates show that, in 2002, the sector
released 36,000 tons of PM10 emissions.

Same as HO1.12

HY1.5: Page 1-11. Figure 1-7 – Emissions as inputs to the model are
baseline uncontrolled emissions at baseline levels of production.

To be addressed in the updated ISIS model/documentation.

KC1.1: Lacking control technologies for CO2, ISIS Cement is not yet
ready to address climate policy except as it relates to electricity and
import prices.…the richness of import-price-related modeling will be
critically important to assessing the cement industry’s response to,
for example, a price on carbon emissions. Both CO2 mitigation technology
and product substitution in demand response are sine qua non for
modeling the cement industry’s response to climate legislation.

To be addressed as part of enhancement of the model.  We are working to
include CO2 mitigation technology and product substitution for modeling
the cement industry’s response to climate legislation. 

KC1.2: How substitutable are the different varieties of Portland cement
with each other and with masonry cement? How well do prices and demands
for different varieties correlate? What about other substitutes for
cement? 

From Larry Sorrels: Gray Portland cement is the most commonly supplied
type of cement because of the wide variety of applications to which it
can be put.  Gray cement held a 77% share of US consumption in 2005
(from Regulatory Impact Analysis for proposed Portland cement NESHAP,
2009).  Sulfate resisting or what is known as Title V cement is the
second largest concrete type at 15% of US consumption in 2005.

SK1.1: A model of the domestic industrial sector that encompasses a
comprehensive approach to emission of pollutants for all three media
(air, water, and solids) would provide significantly greater value than
a more limited focus. 

To be addressed in the updated ISIS model/documentation.

SK1.2: Are there any ways to add CO2 capture options?…incorporation of
domestic and international (CO2) offsets are going to be really
important to include.

To be addressed as part of enhancement of the model.  We are working to
include CO2 mitigation technology and product substitution for modeling
the cement industry’s response to climate legislation. 

SK1.3: Page 1-5, Table 1-1: Use of two sets of numbers for kiln specific
heat input (or energy intensity) in this table and Table 2-4 is
confusing. An explanation of the difference would be helpful.

Table 2-4 has been modified and is now consistent with Table 1-1.

Reponses to Comments on Chapter 2:

Data Used in ISIS

DB2.1: Are transport-related emissions included in the emissions
calculations?…transport of fuel and material to the plant, and
transport of cement to market.…it wasn’t obvious to me whether and
how transportation costs are included in the model. Are the costs of all
materials and fuels the delivered costs?…it is not described how are
costs of transporting product between regions accounted for…

Currently transport related emissions are not included in the mode.
Subject, to availability of data, we will consider inclusion of such
emissions in a future version of the model.

DB2.2: What assumption can be made or is made about the location of the
new unit and the expanded unit in each region? Are all input costs
uniform for all plants within a region?…model user will have to have
information about which plant within a region expands, or where new
plants are located. A reasonable assumption may be to distribute new
capacity proportionately at the location of existing facilities. 

We have incorporated this suggestion in the current version of the
model.

DB2.3: On Page 2-7, bottom paragraph, I had a hard time understanding
the algorithm described. 

Description of SO2 emissions to be expanded in the updated ISIS’
documentation

DB2.4: Page 2-23. The elasticity value cited is -0.88. Is this to be
interpreted as a short run or long run elasticity? Is there reason to
think they differ, or not? Are there long run substitutes for cement?

The value of -0.88 is a short-run price elasticity of demand estimated
by RTI as part of econometric modeling done for the portland cement
NESHAP promulgated in 1998.   Given that there are very few substitutes
for cement, there are not likely to be long-run substitutes for cement.
Further, the user can choose to do a sensitivity analysis by varying the
elasticity values, as elasticity is a user-input that can be specified
in the workbook. 

DB2.5: Page 2-2. “The ISIS model determines the optimal fuel
type…” Does the model account for long run contracts for fuel, or
are all projected fuel choices assuming a spot market for fuel?

The model takes into account regional prices of the fuel as provided by
the Energy Information Administration, and their air pollution impacts,
to decide the optimal fuel-type for a given kiln.  

DB2.6: Page 2-5. “In ISIS, projected <delete: kiln> retirements of
existing kilns…”

Editorial comment, done

DB2.7: Page 2-7, Table 2-5. Explain sources in this table and why they
diverge.

Table 2-5 was modified and numbers are given in terms of lb NOx/ton of
clinker and in terms of lb NOx/MMBtu.  The source is 2007 ACT document.

DB2.8: Page 2-7, bottom. “…location (i.e. feedstock) appears…”
Does this mean to say: “location and the associated feedstock (fuel
type) appears…”?

Feedstock refers primarily to the raw material- limestone. Text
modified.

DB2.9: Page 2-8, Table 2-6. The colors do not appear consistent. For
example, why does Birmingham, with SO2 emissions of 0.09 for the PC
technology, have a white color in the cell?

Moved to “Reponses to Comments on Appendix A.”

DB2.10: Page 2-10. “By-product benefits” Also costs?

Moved to “Reponses to Comments on Appendix A.”

DBS2.11: Page 2-10, Table 2-8.…Is there no (Low NOx burners) penalty
in a kiln?…there is no CO2 penalty identified with the use of a Wet
Scrubber, but later on Page 3-5, a penalty associated with “generation
of CO2 in a wet SO2 scrubber” is described.

Table 2-8 modified to reflect slight CO2 increases for Low NOx burner
and wet scrubber.

DB2.12: Table 2-8 also seems inconsistent with the memo from Andover
Technology Partners with respect to tires and the CO2 implications. 

The following footnote was added to Table 2-8:

“Tires are made of biomass, so there is an incremental CO2 emission
benefit. Because of the high heating value of tires compared to commonly
used fuels, CO2 emissions are lower. Tires produce slightly less CO2
than coal.”

DB2.13: Page 2-12, Table 2-10. It may be useful to add an escalation
factor for construction materials. 

Escalation factor of 5.16% for construction materials was added in Table
2-10.

FF2.1: Demand projections -          In ARB’s draft calculations for
California’s cement industry, ARB staff has assumed a 2% annual growth
rate in cement demand.  Due to the economic downturn, staff will be
reevaluating the annual growth rate.

To be addressed in the updated ISIS model/documentation.

FF2.2: Production and control costs and associated escalation rates -   
ARB has limited information in this area and can not provide comments.

Thank you.

FF2.3: Treatment of imports -       California cement facilities will be
part of a cap and trade system to reduce greenhouse gas (GHG) emissions.
 Details of the cap and trade system are under regulatory development. 
In addition, ARB staff is considering the development of a regulation to
establish blending requirements of supplementary cementitous materials
at concrete batch plants.  The treatment of imported cements will be
considered as part of that effort.

To be addressed in the updated ISIS model/documentation.

HO2.1: In many places, you have used PCA data instead of the USGS
data.…Any State-level (and USA overall) consumption or production data
published by the PCA are taken from USGS data and you should cite the
USGS for such data, not the PCA.

Generally, in this document USGS sources are used for historical data
and PCA sources are used for projections.

HO2.2: Somewhere, you should discuss the issue of data accuracy. In many
places, data ought to be rounded. Few data reported are good to better
than +/- 1%.

Data was taken from different sources. Some data were measured while
some were calculated.

HO2.3: The USGS, not the PCA, should be cited as the source of the value
data and the consumption numbers. For 2005, portland cement sales to
final customers (these are the data that the PCA calls “consumption”
were worth $11 billion; masonry cement sales $0.68 billion; total about
$12 billion.

This is comment to Chapter 1. Introduction. Numbers were changed to
address $12 billion market.

HO2.4: Cite the USGS also for the statement that the top 10 cement
companies had 80% of the production; ditto for all production by State
data.

Moved to Editorial Comments (HOED2.1)

HO2.5: References:…USGS Mineral Commodities Summaries cited—these
are “quick & dirty” summaries that are meant to be just that. We
prefer, where possible, that data be obtained from the Minerals
Yearbook, as this is a MUCH more authoritative publication.

Minerals Yearbook data was used Reference corrected to USGS, 2007a.

HO2.6: Section 2.2.1.1: The USGS has a kiln count of 183, not 181 (=PCA
number?)—but the difference is likely kilns active during year (183)
vs still active at yearend (181).

We use PCA kiln count. The difference (183 vs. 181) is due to the
difference of “active” kiln definition.

HO2.7: Figure 2.1:…there are some very strange groupings re. the
cement plants. For example: Why on Earth is Ash Grove, Leamington, UT
grouped into Phoenix, but Holcim Devil’s Slide is Salt Lake City? Let
me assure you that the Leamington plant feeds markets primarily in UT
(incl. SLC) and NV (Las Vegas) (plus v. minor tonnage, only, into CO and
NM; nothing into AZ). In Las Vegas it competes with the S-CA producers.
Why is Ash Grove, Durkee, OR grouped with Salt Lake City—its market is
primarily OR, WA, ID.…

In ISIS markets are not isolated regions, we have created an extensive
transportation matrix, allowing for inter-market trades.  For example is
known that some plants straddle more than one market: Lafarge’s Sugar
Creek, MO plant can indeed feed westward into the Kansas City area, but
it also ships a lot of cement down the Missouri River (i.e., eastward).
Buzzi’s Cape Girardeau plant probably has the widest geographic
dispersal of its cement than any other plant. Overall, there are all
sorts of flows up and down, in and around, the entire
Mississippi-Missouri-Ohio River system.

To be addressed in the updated ISIS model/documentation.

HO2.8: Section 2.2.1.3: The 2005 demand for portland in 2005 was 122.4
Mt (million metric tons), and for portland + masonry, it was 127.9 Mt.
Thus I question the 127.6 portland number in your report.…PCA demand
forecasts…change…at least quarterly.…it is likely that you have
used reports predating the recent collapse of the economy…The PCA has
revised its forecasts downwards.…we have had plant closures in 2008 (I
think 3) and announcements of about 7 more in 2009,…

The reason for the “127.6 number” is that Hawaii and Alaska are not
included in Table 2-2.

HO2.9: Table 2-3: put the data back into mt. Data include clinker. These
are U.S. Census trade data (the USGS just compiles them into convenient
tables).

Data in Table 2-3 is now in metric tons.

HO2.10: Page 2-5: Cement plants can certainly be older than 50
years!…. Generally, you wouldn’t build a plant on a site having <
100 years of limestone reserves. So the age can be a “site age” or
an “equipment age”.

Language changed to reflect that plants may be 50 years old or more.

HO2.11: The phrase “…ISIS endogenous capacity growth can also lead
to endogenous retirement of kilns.” conjures up all sorts of imagery
unrelated to cement…Couldn’t you just say something like:
“capacity expansion projects can include the retirement of existing
kilns…”?

Language changed to describe that capacity growth may lead to retirement
of kilns.

HO2.12: Page 2-6: “…such as coke…” Be sure specify petroleum
coke (petcoke). Our old tables showed for many years just “coke”,
then we attempted to differentiate petcoke from metallurgical coke
(i.e., from coal) without much success (not sure the respondents
understood the difference). Likely that all coke use (outside of major
blast-furnace/steel complex areas) has been of petcoke, and I am not
sure that metallurgical coke has ever been in common use even in the
steel areas—it’s too expensive.

Footnote added to explain that PCA does not specify if “coke” is
metallurgical coke or petroleum coke. Authors believe it is the latter.

HO2.13: Figure 2-2: awfully crude breakout of fuels.…Also, do the
percentage data refer to tonnages or to the heat contribution (Btu)? If
I use actual industry-reported heats, my own data for 2005 show: Natural
gas (3.72% total Btu); fuel oil (0.76% [seems low]); coal (64.1%);
petcoke (19.3%); metallurgical coke (nil); tires (3.2%), other solid
wastes (0.5%); liquid wastes (8.4%). Your report cites total heat (EIA)
of 451.2 trillion Btu. My (USGS) data for 2005 has 366.7 trillion Btu
excluding electricity,…Using actual data, if I include the
electricity, then we jump to 413.4 trillion Btu. Thus, I suspect that
the EIA data are either too high or that they, at the very least,
include electricity. Based on the rest of your report,…skip the
electricity (in your heat analysis).

New pie chart inserted utilizing numbers given by the reviewer.

HO2.14: Table 2-4: shouldn’t these data match those of Table 1-1 (or
vice versa)?…the PCA data on unit consumption of heat…generally are
based on an “equivalent ton” which is a bit of an apples & oranges
mix.…You can recalculate to a per-ton-of-clinker basis, but I do not
know if you have done this in Table 2-4. Also, I don’t recall the
PCA’s Labor & Energy surveys mentioning gas flows—so it is unclear
if the NM3/kg data are from the PCA or some other source.…you should
cite the actual page(s) from which the data are taken.

Table 2-4 has been revised to match data in Table 1-1.

HO2.15: Page 2-8. Your calcination emissions factor of 0.53 t CO2 per
ton of clinker is slightly too high. Instead of adopting this Andover
Technology default…you should use the IPCC default (which is what the
EPA usues for the USA GHG Inventory.…So use 0.51 or 0.52 (the latter
includes a 2% addition for “lost” [not returned to kiln] CKD).

Emission factor of 0.52 taken

HO2.16: Table 2-7: I’m not sure you’ve made clear why you are
showing H2O production. 

H2O is product of combustion of fuels.

HO2.17: Table 2-8: low-NOx burners are great,…, but lots of plants
report…lower clinker output when using these burners.

Low NOx burners control fuel and air mixing and -as a result- reduce
peak flame temperature and form less NOx. The longer, less-intense
flames resulting from lower temperature reduce thermal NOx formation by
about 30%. Low NOx burners may increase the amount of un-burnt carbon. 
By reducing excess air from 10 to 5%, NOx emissions can be reduced by
approximately 15%. Reducing excess air levels results in indirect
reduction of NOx emissions per amount of clinker produced (ACT 2008).

HY2.1: Page 2-6. Table 2-4: Spell acronyms in the table (SFC, EGFW) to
be consistent with the table in the appendix. The EGFW in Nm3/Kg for dry
seems low. Hector will share Holcim’s calculation method at 10% oxygen
to recalculate the EGFW.

Acronyms spelled out.

HY2.2: Page 2-7. Table 2-5 Update with the most recent ACT data (2008
ACT) and show data in lb/ton clinker

Table 2-5 was updated.

HY2.3: Page 2-9. Table 2-7 Label symmetrically lbmolesH2O/MMBtu, H2O is
missing in some places

Done

HY2.4: Page 2-10. Table 2-8 Are CO2 reductions total CO2 (process and
fuel)? If so, express it to be total CO2. Are numbers absolute #s or
intensities? Can sewage impacts be included? PCA has given EPA a
workbook with the fuel and power use for the cement industry. This
workbook can be used to compare #s. The workbook is part of the WBCSD
initiative. Hector will find out who at EPA PCA sent the workbook data. 

Numerical values of CO2 emission reductions (intensities) are for fuel
use, as tires produce slightly less CO2 than coal. Sewage impacts are
not included in CO2 emission reduction.

HY2.5: Page 2-11. Table 2-9 Are these #s fuel intensity or absolute? We
might think that the #s are intensity related not absolute. When going
from a wet to a PC you have more capacity, you produce more and
therefore you will have more emissions in the absolute no matter that
the process is more efficient.

These numbers are intensities.

HY2.6: Page 2-12. Table 2-10 Specify in the model that these parameters
are to be chosen by the user.

Comment added.

KC2.1: The documentation (Section 2.2.3.2) mentions a cement demand
elasticity of -0.88,…model results show no changes in demand in the
policy case, despite significant allowance prices. Is this a disabled
feature? 

Demand response in the model can be enabled by choosing value of
“Demand RESP” to be 1.

KC2.2: The rationale behind the choice of region definitions needs more
documentation…assuming uniform region-wide parameters could
systematically underestimate the potential competition and exchange
between neighboring plants in different regions.…it would be nice to
know or estimate how much of each plant’s commerce is within its ISIS
region versus outside of it.

The model outputs include quantity of cement traded-in and traded-out by
each of the regions. 

KC2.3: Uncontrolled NOX emissions intensities vary…by a factor
of…Have you examined the sensitivity of results to these
parameters?…how much investment in monitoring infrastructure would
kilns currently need to comply with either cap-and-trade or rate-based
policies?

To be addressed in the updated ISIS’ documentation.

KC2.4: OMB’s 7% discount rate makes sense for analyzing the costs and
benefits of policy results from ISIS, but it is not a basis for
reflecting the discount rate cement companies use…

Discount rate is user-defined parameter and can be changed.

KC2.5:…model’s energy prices escalation factors reconcile with those
in AEO 2008 Table A1…, they disagree with those in Table A-3 (by
sector),…more recent projections by EIA and IPM have much higher
future price increases (2–3% annualized) than AEO 2008.

This is a user-defined parameter and can be changed

KC2.6: Given transportation costs and the fact that 10 companies account
for 80% of production overall, is market power an issue in any regions
or subregions?

Discussions with the industry indicated the industry believes it
operates in a perfectly competitive market and, therefore, there are no
issues related to market power.  In addition, historical demand exceeds
capacity.

Instead of treating markets as isolated regions, we created an extensive
transportation matrix, allowing inter-market trades possible (which is
more realistic) thereby reducing the possibility of market power.  Also,
we took into account comments inf “EPA’s Draft Economic Analysis of
Air Pollution Regulations for the Portland Cement Industry (APCA,
January 1997) regarding choice of appropriate market structure for
cement manufacturing.

KC2.7: Are PCA estimates…corroborated by independent sources?…is the
PCA still projecting 27 new kilns through 2010?

PCA is currently reviewing the ISIS’ model, we expect to receive
updated information of projections.  Currently, EPA is using for the
Portland cement NESHAP projections of 20 new kilns in 5 years from their
base year (2005).

KC2.8: In the real world, do producers buy at a world market price or
benefit from multinational ownership?

There is no “world market” price. Prices are developed regionally. 
Producers can benefit from multinational ownership by participation in
imports.

SK2.1: How are the regions modeled for cement going to map into other
industries…? Integration with other industries is going to be really
vital…especially…if costs are higher for one industry than another
to control certain pollutants.

Regions for one-industry may be entirely different from regions in
another industry. ISIS-cement has regions pertinent to cement industry
only. Future development on the ISIS framework including other
industrial sub-sectors may take this into account.

SK2.2: The model seems to rely heavily on imports…to meet emissions
requirements.…is there a cost curve that increases as more imported
cement is demanded?…(how were) import levels in future years
(estimated)? Does (model) proportion future imports the same way they
were divided in 2005?

Model proportions future imports based on proportions in 2005. However,
these proportions can be modified by the modeler based on additional
information, if available. 

SK2.3: Interregional trade seems like it would be very sensitive to fuel
prices – using diesel prices avg. 1994-2005 may be misleadingly low.

Yes, inter-regional trade is very sensitive to transportation costs, and
transportation costs are adjusted for 2005 based on increase in the
average price of diesel, which is the main fuel used for trucking
cement.

SK2.4: Page 2-7; Section 2.2.1.8 Emissions: This section only discusses
air emissions. A discussion of all substances emitted to all media would
provide greater value.

To be addressed in the updated ISIS model/documentation.

SK2.5: Page 2-2, First Paragraph: It is stated that the ISIS model
determines the optimal fuel type. What is the basis of this selection?
Is there any information available for the existing fuel type, e.g., for
the base case? Can the fuel type be an input to the model?

The basis of selection is relative fuel-prices in a given region and
emission characteristics of various fuels. Yes, fuel-type can be an
input to the model, but prescribing a fuel-type might restrict the
model’s ability to switch fuels in response to a policy.

SK2.6: Page 2-7, Section 2.2.1.8: In the first paragraph, it is stated
that ISIS model uses emission intensities to estimate NOx, SO2, and CO2
emission projections. However, from the model input file…; it is not
obvious if this has been done with the CO2 emissions. It appears that
CO2 emissions for the year 2005 are calculated by multiplying the
clinker yearly production rate with the same factor (0.45 for fuel) for
all kilns. Also, the 2005 baseline figures for NOx and SO2 emissions are
estimated based on a constant multiplier for all kilns. Notes 10, 11,
12, and 14 are not provided, which may have explained the reasons.

To be addressed in the updated ISIS model/documentation.

SK2.7: Page 2-9, Table 2-7: This table lists different CO2 emission
rates for PRB and bituminous coals…However, in the model input file,
only one type of coal…is used. It is assumed that the model has the
capability to use different coal types…

Yes, model has this capacity.

SK2.8: Page 2-11, Section 2.2.3.2:…what is the basis for (the overall
economic analysis) these calculations: current or constant dollars? How
are the construction times for the controls taken into account? Does the
model estimate some of the construction costs (e.g., allowance for funds
used for construction),…? In Table 2-10, what does the annual
escalation rate of 3.09% for the variable O&M cover – all variable
costs or some of them? If all, then why different factors for limestone
and gypsum? Is there a separate factor for lime as well?

To be addressed in the updated ISIS model/documentation.

All costs are represented in constant (2005) dollars. Different factors
for limestone and gypsum were obtained from historical price data
published.

Reponses to Comments on Chapter 3:

ISIS Mathematical Framework

DB3.1: One could move the Objective function in Section 3.6 up to the
beginning of Chapter 3. I would create a table…Then, the specifics of
the algorithm can be explained in the other parts of Chapter 3.

To be addressed in the updated ISIS model/documentation.

DB3.2: Accounting of capital cost is not transparent…It is not
apparent whether the variable cost in year 10 includes capital cost for
the newly built capacity (long run marginal cost). Perhaps the more
appropriate question is whether annual fixed costs is included in the
calculation of market prices. Is annual fixed cost considered a variable
cost when the model is solving in annual time steps? 

To be addressed in the updated ISIS model/documentation.

DB3.3: Page 3-2.…What is the relationship between the chosen values
(of the constraints limiting capacity changes) and historical
experience? 

To be addressed in the updated ISIS model/documentation.

DB3.4:…on Page 3-3…imports are constrained to historic experience.
However, many are concerned that under strict domestic climate policy
there will be infusions of imports from countries that have not take on
climate-related obligations…At least, this constraint should be
tagged…

To be addressed in the updated ISIS model/documentation.

DB3.5: Page 3-2, middle. I don’t see how the constraints relate. 

To be addressed in the updated ISIS model/documentation.

DB3.6: Bottom. “Note that replacements are associated with individual
units….expansion…changes are…in regional capacity.” Then why are
they summed?

To be addressed in the updated ISIS model/documentation.

DB3.7: Page 3-5. What is “priceration” in Equation 3.2.5?

To be addressed in the updated ISIS model/documentation.

DB3.8: Page 3-5, middle. Energydisl…and polbasesubfuel…Are these the
same thing? If not, what is the difference?

To be addressed in the updated ISIS model/documentation.

DB3.9: Equation 3.3.2. What is “cp”?

To be addressed in the updated ISIS model/documentation.

DB3.10: Page 3-7.…Can some narrative be added? Why is the sum in
Equation 3.3.9 divided by the number of pollutants? What are units
throughout these equations?

To be addressed in the updated ISIS model/documentation.

DB3.11: Page 3-9.…“catconsumptcost”…is not defined as a price
times a quantity. However, see Equation 3.4.11a…cost is defined as a
price times a quantity. 

To be addressed in the updated ISIS model/documentation.

HO3.1: Page 3-2: the factor: produ_r(t,r)/0.92 has a clinker ratio
(clinker/portland) of 0.92. This is too low for a good average. Even
including the blended cements and masonry cements, the USGS data show
the ratio to be right on at 0.946-0.950…. Anyway, 0.92 implies 8%
gypsum and/or other additions, and this is too high right now. We can
approach a 92% clinker factor hypothetically if we assumed that we had a
95% market share for true portland cement @ 95% clinker + 5% gypsum, and
5% market share of masonry cement averaging 50% clinker (this is
probably a bit low): (0.95 x 0.95) + (0.05 x 0.5) = 0.9275.

To be addressed in the updated ISIS model/documentation.

HO3.2: Again, the import data that you are using appear to be for both
cement and clinker. The factor imports (t,r) is likely prone to large
error because we simply do not know how much cement coming into a
certain Customs District is consumed in that district.….

To be addressed in the updated ISIS model/documentation.

HO3.3: Production and Capacity Changes: the lead-in sentence is a bit
unclear: “Regional production can be from existing kilns, kilns added
at a plant (i.e., expansion kilns), newer kilns replacing kilns at a
plant (i.e., replacement kilns), and new kilns.” I think you mean to
say something like: “…Future production…existing kilns, upgrades
of existing kilns (debottlenecking, major technological upgrades), new
kilns at existing plants (perhaps replacing existing kilns), and new
plants altogether.”

It is unclear if your discussion, and projected capacity changes, fully
account for plant shutdowns…. Have capacities been properly/uniformly
defined for all new kilns (are they summed on the same basis)?

To be addressed in the updated ISIS model/documentation.

HO3.4: Pages 3-2 through 3-3: Equation 3.1.3. for regional demand: I
suggest using the word “flows” instead of “trade” when referring
to inter-State or inter-market movements of domestic cement.

To be addressed in the updated ISIS model/documentation.

HO3.5: Equation 3.1.4 only holds if the import: total demand ratio is
accurate, and I’m not sure this is the case.…How are you determining
where cement imported at location X is being consumed?

To be addressed in the updated ISIS model/documentation.

HO3.6: Page 3-3: the price data for imports (Equation 3.2.5.) are very
weak. You only have import values reported on a CIF basis (USGS monthly
and annual reports) and Customs Value (USGS annual reports). The CIF
value does not represent the actual price because it has not yet been
offloaded from the ship: that is, the true price is CIF + various port
charges + markup.…High import ratios exist for Florida, Los Angeles,
San Francisco, San Antonio—yet the Table (4.1) designates (**) only
Florida and Seattle as having high ratios.…The PCA may have predicted
higher import ratios in the future, but they certainly have fallen in
2007-09 and likely will stay down in 2010 as well.…

To be addressed in the updated ISIS model/documentation.

HY3.1: Comments on Spreadsheet. Add plot on retirement, new capacity,
etc

To be addressed in the updated ISIS model/documentation.

KC3.1: Equation 3.2.1 appears to set annual fixed costs proportional to
a unit’s production rather than its capacity. If I am interpreting
this correctly, it is a serious conceptual mistake.

Addressed by adopting standard new kiln capacity of 1.2 million tons

SK3.1: Chapter 3, Can the model accommodate different safety valves
built into a policy framework. Can a case for cap-and-trade with a limit
on allowance price (with other required caveats) be run?

To be addressed in the updated ISIS model/documentation.

SK3.2: Cell F6 should have some sort of label telling what it is.

To be addressed in the updated ISIS model/documentation.

SK3.3: Fuel savings (cell J4) was not able to generate any scenarios
where that was populated - what is it?

To be addressed in the updated ISIS model/documentation.

SK3.4: Rows 9-24 would be more helpful if the formatting had commas,
truncated decimals to one or two places.

To be addressed in the updated ISIS model/documentation.

SK3.5: What does “eps” mean? 

To be addressed in the updated ISIS model/documentation.

SK3.6: I ran a CO2 reduction case but no allowance price was calculated
for CO2 in the Output spreadsheet – I think this would be a useful
output.

To be addressed in the updated ISIS model/documentation.

SK3.7: Cells AL30:45 – these are not labeled, and it’s not clear
what this is.

To be addressed in the updated ISIS model/documentation.

SK3.8: Rows 53-68 and 72-87 what are the units? Any significance for the
red font?

To be addressed in the updated ISIS model/documentation.

SK3.9: In the CO2 case I ran…ISIS expanded capacity…and built…new
capacity and seemed to keep all existing capacity,…why would the model
build new capacity and not use it if it is not retiring older dirtier
capacity?

To be addressed in the updated ISIS model/documentation.

SK3.10: Cell C2 – do you mean Newark, DE?

To be addressed in the updated ISIS model/documentation.

SK3.11: Would be helpful to be able to adjust diesel prices – using
AEO2005 may not be the most reliable data.

To be addressed in the updated ISIS model/documentation.

SK3.12:…SO2 and NOx have lots of data on their emissions factors, but
CO2 does not. The emissions factors for CO2 (rows 36-38) seem lower than
those in EIA’s Table 6-1…

To be addressed in the updated ISIS model/documentation.

SK3.13: For the % reduction policy options inputs (cells F5:H22), what
if the % reduction compared to? Is it a year-by-year?…

To be addressed in the updated ISIS model/documentation.

SK3.14: Another useful input would be to be able to manually enter an
emissions cap…

To be addressed in the updated ISIS model/documentation.

SK3.15: K5:M5: don’t understand “miner” in these cells.

This parameter is added to be able to specify a minimum emission
reduction requirement by each production unit. 

Reponses to Comments on Chapter 4:

Calibration of Cement ISIS Model

DB4.1: It would be helpful in Chapter 4 if an explicit equation showed
the variable cost calculation with the calibrator constant included…

To be addressed in the updated ISIS model/documentation.

DB4.2: Page 4-4, middle…Why is a single calibration constant used to
adjust variable cost of production? 

To be addressed as part of the enhancement of the model.  We are working
on developing a more general, spatial equilibrium version of the model
that will address this issue.

To be addressed in the updated ISIS model/documentation.

DB4.3: Page 4-4, top. “the marginal <?> of the regional…”

To be addressed in the updated ISIS model/documentation.

HO4.1: Section 4.1.3.1: I am not sure that the assumption that
Priceimports = producer price is at all valid.……What is the
percentage criteria for this designation (import-dominated markets)?…

We calibrated import prices for each demand region to arrive at
historical import levels supported by data.

HO4.2: Table 4-1: why 2004 data and not 2005? Again, switch to metric
tons! What is the basis of the production data in Table 4-1?….

To be addressed in the updated ISIS model/documentation.

HO4.3: Table 4-2: where are you getting your price data; and why are you
showing 2003-2004 and not 2005? Need to cite a data source.

To be addressed in the updated ISIS model/documentation.

HO4.4: Table 4-3: The emissions shown are too precise.

To be addressed in the updated ISIS model/documentation.

KC4.1: Estimating and predicting based on a range of years rather than
just one for each would improve the calibration procedure. What makes
±10% a reasonable range for predicting values one year later…?

To be addressed in the updated ISIS model/documentation.

KC4.2: Variable cost seems like an important parameter to be imputing a
constant adjustment from a regression over (it appears) two years of
data. How sensitive are the results to the choice of this constant (or
the choice of using a constant)?…my own limited investigations…show
mostly level shifts.

To be addressed in the updated ISIS model/documentation.

KC4.3: Constant escalation factors for fuel prices seem too
simplistic…Ideally, a supply curve informs (of) the cost of fuels.
Absent that, ISIS should use actual annual or multi-annual projections
rather than an annualized 20+-year average.

To be addressed in the updated ISIS model/documentation.

KC4.4: Short of process changes like CEMStar, are marginal changes in
feedstocks available to kilns in the real world? How much difference
does this make?

To be addressed in the updated ISIS model/documentation.

SK4.1: Is there any plan for showing the indirect effect on air
emissions of changes in the electricity consumption caused by
modifications required by policy decisions at cement plants? 

To be addressed in the updated ISIS model/documentation.

Reponses to Comments on Chapter 5:

Illustrative Analysis

DB5.1: What constraints affect the choice of technologies for pollution
reduction? Can only one technology be chosen? Can it be changed? This
appears to be happening in Figure 5-4 on Page 5-4…the rise and fall of
LNB+CEMStar. When this can happen, how are capital costs considered? Are
opportunity costs properly accounting for only marginal going forward
costs (avoidable costs)? 

To be addressed in the updated ISIS model/documentation.

HO5.1: Figure 5-1 and the discussion of it lead to the conclusion that
the NOx cap may be too restrictive.…the emissions from a given plant
typically swing rather widely on a minute-by-minute basis.…it has
always been an issue as to how one is to determine the actual emissions
volumes over a specified period, as it could affect whether a plant
exceeds its NOx cap or not. I think this report needs to discuss this.
In contrast, Figure 5-2 would seem to illustrate the relative ease in
controlling SOx.

To be addressed in the updated ISIS model/documentation.

HO5.2: Figure 5-3:…it is confusing why the emissions caps and banking
are not shown…I do not know if the projections in Figure 5-3 have made
allowance for clinker imports and changes therein. Given a likelier
harsher reality re. plant closures, these projections may need to be
redone.

To be addressed in the updated ISIS model/documentation.

HO5.3: One important issue with CemStar is that steel slag may not be
available locally.

To be addressed in the updated ISIS model/documentation.

HO5.4: Figure 5-4…why there is such a huge drop in LNB + TDF in 2011
wrt. 2010. Are the subsequent (2011--) years in addition to the 2010
levels? (i.e., are they just shown as margins?). Was there a scale
change post-2010?

To be addressed in the updated ISIS model/documentation.

HO5.5: Section 5.1.3: The total demand in the near term has now likely
been revised downwards by the PCA.

To be addressed in the updated ISIS model/documentation.

HO5.6: Figure 5-5:…why the red dot has been displaced rightwards…and
what happened to the red dot for 2025?…I do not see much difference
between the base case and the policy levels.

To be addressed in the updated ISIS model/documentation.

HO5.7: Figure 5-6: Unless I am missing something, the blue dots are not
needed at all—the sum is always the top of a stacked bar graph.…I do
not understand how you have determined the (foreign) imports into the
various markets…So this figure needs some work….

To be addressed in the updated ISIS model/documentation.

HO5.8: Section 5.1.4…the cost analysis seems to show that, of the
total increase in costs ($1.58 billion), all but $0.01 billion is due to
control technology. What about increased fuel costs?

To be addressed in the updated ISIS model/documentation.

HO5.9: Basically OK. Perhaps you need to again mention here (Summary)
why CO2 emissions could drop under SO2-reduction strategies…

To be addressed in the updated ISIS model/documentation.

HY5.1: Page 5-6. Table 5-1 control cost #’s need to be updated to
differentiate between new capacity in the business-as-usual vs. policy
scenario.

To be addressed in the updated ISIS model/documentation.

SK5.1: An ability to store multiple scenario cases would be good. 

To be addressed in the updated ISIS model/documentation.

SK5.2: Page 5-1; Section 5.1.1: The emissions cap and trade example
provided is overly simplistic.…all banked allowances are consumed by
the end of the second phase of the rule…banking of allowances would
most likely continue beyond the end of the forecast horizon being
modeled. 

To be addressed in the updated ISIS model/documentation.

SK5.3: Page 5-5, Figure 5-5: Can there be a scenario, where the price of
cement is affected to the point that it starts to affect demand? 

To be addressed in the updated ISIS model/documentation.

SK5.4: Page 5-6 Table 5-1: The increase in the combined import and
control costs ($2 billions) does not match the increase in the total
cost ($1.58 billions) for the example case.

To be addressed in the updated ISIS model/documentation.

SK5.5: General Comment: The costs associated with the base and policy
cases should be reported in terms of factors (e.g., $/ton of clinker or
$/ton of air pollutant removed) that can easily be understood and
compared to the effects of other similar policies.

To be addressed in the updated ISIS model/documentation.

Reponses to Comments on Chapter 6:

Summary

HO6.1: The Summary seems to lack discussion of CO2—you need to at
least mention it (and then you can say that your cost modeling was
confined to NOx and SOx.

To be addressed in the updated ISIS model/documentation.

Reponses to Comments on Appendix A

All comments to be addressed in the updated ISIS model/documentation.

DB2.9: Page 2-8, Table 2-6. The colors do not appear consistent. For
example, why does Birmingham, with SO2 emissions of 0.09 for the PC
technology, have a white color in the cell?

DB2.10: Page 2-10. “By-product benefits” Also costs?

HOA.1: Table 1: the exit flow gas rates would show a lot of variation;
not sure how “good” an average these rates are. By the way, Bhatty
et al. (PCA 2004) cited is a huge volume, so a page reference would have
been helpful.

HOA.2: Table 2: although the heat values shown are reasonable, it should
be pointed out that, in reality, most fuel categories have a range of
heat values,…I would also comment that the values in Table 2 are
expressed with more precision than is warranted.…Table 2 appears to be
at the upper end of the range.

HOA.3:…I do not like the default…(0.53 t CO2/ t clinker) because it
is based on an average CaO content of clinker of 67.6%. This is too
high—…EPA in the USA GHG emissions inventory uses the IPCC default
(w/ CKD) of 0.52 and I would suggest that the ISIS model do the same.

HOA.4: Page 8:…I agree with Staudt that the NOx data show a lot of
scatter. I would note that his observation that the larger kilns are
more efficient (lower NOx) is perhaps not supported by his graph—yes
it shows this, but there are only 3 data points for the kilns > 1 Mt/yr
capacity. Is this statistically valid?

HOA.5: Figures 5a-5c: Author is being “kind” in his characterization
of the data: the correlations would appear to be not statistically
significant.…

HOA.6: Page 15: perhaps this is a semantic point, but Dragon Cement’s
switch to dry technology involved shortening the existing kiln tube, not
by bringing in a new tube.

HOA.7: While it is true that there is regional variation in the sulfur
content of “limestone” (and the sulfur in “limestone” is from 3
main sources: sulfide minerals such as pyrite; kerogens; and
gypsum/anhydrite.), there is as much, or greater variation from the
S-contribution from the coals: there is a major difference in the
S-content of coals (S in coal is from kerogens and sulfides like pyrite
and marcasite).

HOA.8: The sulfur content of the “limestone…would not normally be a
criteria for selection of a new quarry.

HOA.9: Given the small number of kilns in most of the markets, I am not
sure that one ought to read too much in the mean vs. median emissions
data in Tables 8 and 9. 

HOA.10: Regarding the use of tires,…I am not sure that I agree that
tires yield slightly less CO2 than coal: perhaps on a dry weight basis
they yield slightly more. Depends on how you relate the emissions: per
weight of fuel, per ton of clinker…

HOA.11: I agree that tires are not fossil fuel, per se, but most involve
a considerable input of the products of the petrochemical industry in
their manufacture.

HOA.12: Page 13: Preheaters and precalciners are described as
“post-kiln” combustion technologies.…but most people would…call
preheaters and precalciners: “pre-kiln” technologies (that’s why
“pre-“ is in their name!). I think most calciners are designed to
burn about 60% of the (total kiln line) fuel, not 50%; though perhaps
the oft’ quoted 60% includes riser duct firing.

HOA.13: Page 26: LWS scrubber: basically fine. Here and elsewhere, use
calcite, or calcium carbonate, instead of limestone. Limestone is a
rock, only very rarely is it pure calcium carbonate.

HOA.14: Page E-31: minor issue: gypsum is used to make finished cement,
not finished concrete. Another issue: CaSO4 is the formula for the
mineral anhydrite. Gypsum has the formula: CaSO4∙2H2O .

HOA.15: Page 30: CKD is described as being made up of oxides of Ca, K,
and Mg. You’ve left out Si!

HOA.16: Page 30 Paragraph 2 states that CKD disposal is not an issue
with dry-technology kilns. This is not true. Any kiln technology can
burn CKD and any can have a problem burning CKD—it depends primarily
on things like the alkali content of the CKD and the resulting clinker.
If there is a regional ASR problem with the local aggregates (in
concrete) then you need to keep the alkali content of the clinker down.
This may be well-controlled if you have an alkali bypass system, but if
you do not, then your ability to burn CKD may be severely limited.

You can certainly burn CKD in a wet kiln: the slurry viscosity can be an
issue, but you can also add a bit more water. CKD was burned at
Dragon’s wet kiln, for example, but only in relatively modest amounts
(they did not burn all of it).

HOA.17: CemStar: I would add a bullet for the CO2 reduction! On the
negative side, CemStar may add to a plants emissions of Cr+6. I think
this is why Cemex’s Davenport (N-CA) plant is presently idle. Good for
you in getting the $16/ton royalty figure:…The high royalty definitely
limited interest in the technology.…it is likely that future (new)
users of CemStar will be paying <<$16/ton royalty (if any) for the
process. At $16/st, CemStar becomes quite expensive relative to the cost
of the steel furnace slag (a few dollars per ton only, + transportation)
and its handling. I do not understand this $72.59/st cost for clinker
(seems excessively high), nor why it is even relevant.…What you seem
to be doing is pretending that you stick slag into the finish mill to
make cement (that it competes with clinker) when, in fact, it is
“competing” with (some of the) limestone as a feed to the kiln.

What you need to do is to compare the cost of making clinker without
using CemStar vs the cost to make clinker using CemStar. CemStar always
involves incorporating steel slag into the raw mix as a partial (3-10%)
substitute for limestone (etc…). Without CemStar: cost = A (cost to
quarry, comminute, & blend the raw feed materials) plus P (the cost of
pyroprocessing (fuel costs etc…) to make clinker. This is the cost to
make the clinker, period. With CemStar: for every ton of clinker made,
you are quarrying, comminuting, and blending less “normal” feed, so
A is lower by an amount B. But, against this, is the cost (A’) of:
buying the slag, the royalty for the process, crushing the slag to
1-2” size, and blending. So the resulting feed cost becomes (A-B +
A’).

The pyroprocessing cost is the cost of calcining the original feed mix
(in the now smaller amount), plus the cost of melting/disassociating the
slag, plus the cost of sintering. Because there is less carbonate to
calcine (per ton of clinker), the fuel (heat) required for calcination
will be less. (Calcination is where 60% or more of the heat is
consumed). Also, because the slag melts/disassociates easily, you are
getting some of your silica, iron, and aluminum (oxide) for less heat
(and maybe faster) than would be the case burning clays, shale, and
quartz sand. So the cost savings are related to the reduced heat/fuel
charges of the pre-sintering stages of pyroprocessing (P’). For the
sintering part of the clinker-forming process, there probably will not
be any savings. So the pyroprocessing cost becomes P-P’.

So the overall comparison is A + P vs. (A-B + A’ + P – P’). And
the overall savings will be nowhere near $45-50/ton of clinker! 

The reduction in CO2 emissions will be, on a per-ton-of-clinker basis,
will be something like: for calcination emissions: (1-C’)*C, where C
is the calcination CO2 for the carbonates burned in feed that does not
incorporate CemStar, and C’ is the percentage of carbonate
substitution by slag. That is, 5% (C’=0.05) substitution yields:
(1-.05)*C = 0.95C. i.e., this reduction is stoichiometrically
straightforward. The fuel emissions reduction is a bit less so. If F is
the non-CemStar fuel emission for the pre-sintering stages, the new
emission (w/ CemStar) becomes something like: ((1- C’)*F) + G + S
where the first term relates to the reduced heat of calcination and
low-temperature pyroprocessing of the normal feed mix; G is the low
temperature fuel requirements to deal with the pre-sintering
disassociation of the slag; and S is the fuel requirement for sintering.
I am assuming that S does not change (i.e., fuel emissions without
CemStar = F + S). The CemStar savings on fuels stems from the fact that
G << C’*F.

HOA.18: What this all means is that the statement on Page 36 that the
CO2 reduction (overall, on a per ton clinker or cement basis) will be
directly proportional to the percent slag substitution is not quite
safe—it will hold for the calcination part of the emissions, but may
be different for the fuel side of the emissions.

HOA.19: Page 35: Again, if I am reading this correctly, it appears that
a claim is made that CemStar increases the electricity consumption by
the finish mill. Well, if you assume that you get 5% more clinker for a
5% CemStar (slag) introduction, then, yes, it will take 5% more
electricity to grind that extra clinker. But so what? This is like
saying that if you get a salary increase, you will pay more taxes even
though the tax rate doesn’t change. This is not interesting. What
would be interesting is if the use of slag (CemStar) made a clinker
(perhaps a harder clinker) that was costlier to grind (more electricity
per ton). But I do not think that this is the case: the cost (either per
ton of clinker or per ton of cement) to grind the clinker into cement
does not change as a result of using CemStar. There may be some
additional electricity in the raw mill end of the plant to handle the
extra ingredient (slag) but this is likely to be more than offset by the
reduced electricity costs of comminution of the feed: unlike the
limestone and other feeds, the slag does NOT need to be ground, just
coarsely crushed. [the CemStar patent “discovery” was not the
chemical contribution to the clinker-making process. It is the fact that
you don’t need to grind it—only crush it to 1-2” diameter. The
utility of slag had been known for many decades, but it had not been
popular earlier because it was assumed that you had to grind it like
everything else, and slag is indeed hard (costly) to grind].

HOA.20: Page 37: The wording (Conversion from wet to dry) is a bit
imprecise. You can convert a plant from wet to dry by replacing wet
kilns with dry kilns and/or by converting wet kilns to dry technology.
Dragon Cement in Maine was a true kiln conversion: the old wet kiln tube
was retained (but about 1/3 of its length was cut off), and a
preheater/calciner tower was added. Holcim @ Holly Hill, SC was a
replacement of the 2 wet kilns (which were shut down) with a completely
new, unrelated, dry kiln (precalciner kiln). Regarding the 300,000 ton
example—I would only comment that one would never build a kiln of such
small capacity today, nor would it be worthwhile anymore to convert a
300,000 t wet to a 300,000 dry kiln.

HOA.21: Page 38: states: “Water consumption would also drop. Moisture
added to the materials of a wet kiln amount to about 0.75 tons per short
ton of clinker.” This is too high (too much water). Wet and dry kilns
both need about 1.6-1.7 tons of raw materials per ton of clinker.
Typical water content of wet kiln slurries is 35-40% by weight. So a
35-40% water slurry containing 1.6-1.7 t of raw materials (destined to
make 1 ton of clinker) would contain 0.56 – 0.68 t water.

But, of course, water is also used in cement plants as a motor coolant
and in the finish mill (as a spray) to make sure the calcium sulfate in
the cement remains (mostly) gypsum and does not dehydrate to plaster or
to anhydrite.

HOA.22: Page 42: The question is asked why haven’t all the existing
preheater kilns been converted to preheater-precalciner technology?
There are at least 3 answers to this:

if the preheater kiln is small (capacity), the conversion boost in
capacity might be insufficient to justify the cost;

significant projects like this typically require a litmus test of the
plant still having at least 50 years of limestone reserves. 

significant projects like this likely will put the plant into a stricter
regulatory framework re. emissions (the existing plant may have
grandfathered in).

HOA.23: General comment: One significant omission from this
study—unless I missed it somewhere (it is a long report!) are certain
upgrades to capacity that simply involve debottlenecking. This could
involve things like:

upgrade/replace clinker cooler

upgrade raw mill

upgrade finish mill (e.g., install roller mill instead of ball mill—it
will greatly reduce electricity consumption)

Upgrade storage facilities (raw feeds, clinker, cement) and overall
materials handling.

Upgrade dust collectors

HOA.24: Page 44: don’t need to capitalize greenfield(s) and
brownfield(s).

HOA.25: Page 45: Not sure that “attainment” status is clear to all
readers—may be environmental jargon.

HOA.26: Item #2 (reserves). 100 years for a greenfields project is about
right. To this I would again note that major capex projects need c. 50
years of remaining reserves or more (some companies may be satisfied
with 35 years).

HOA.27: Item #3 labor: I do not think labor is important…Look, they
have cement plants all over the world. Labor can always be found. The
more skilled positions will draw engineers (et al.) from far away. When
I have visited cement plants, I am commonly struck by the fact that most
of the workers have been there for ages—they have had or are planning
on having their whole career there. They evidently like the work. The
pay is adequate and steady; it’s relatively clean, and relatively
safe. It’s a more recession-proof industry than most. So there’s not
necessarily a lot of overturn (oddly enough, except for the plant
managers). People learn the work; they get trained.…

HOA.28: You offer the notion that the USGS has/will map out suitable
limestone deposits. Unlikely. Cement companies have their own staff of
geologists, or will hire a consulting geologist, for this purpose (they
may use USGS and/or State geological maps to broadly identify where the
limestone may be found).

HOA.29: Page 46: Decision tree: add: consider adding a terminal (instead
of a plant); consider swap arrangements (to supply cement to customers
outside your own plant’s “range”.)

HOA.30: Round the midpoint costs to 2 significant-figures: e.g., $280
instead of $277.

SKA.1: Page 1, Table 1: It needs to be clarified if the specific fuel
consumption data provided includes only kiln fuel or non-kiln fuel
sources, such as raw material dryers, auxiliary boilers, etc.….

SKA.2: Page 4, Second Paragraph: The figures quoted in this paragraph
are based on pure limestone. It’s not clear if all of the figures in
this paragraph, e.g., for the steelmaking slag, are based on pure
limestone.

SKA.3: Page 6, Total CO2: What is the assumption used in the model with
regards to the cement kiln dust, which is also a source of CO2 emission?

SKA.4: Page 7, Table 4: The range value (1.9-3.1) for the wet kiln NOx
emissions does not cover the average value (6.2).

SKA.5: Page 27, Appendix 4: The heating value of natural gas, 26,037
Btu/lb, shown in the last column is too high. This value is generally
between 21,000 and 22,000 Btu/lb.…

SKA.6:…Many of the cost sources used in this document are from years
prior to 2005 and they have been adjusted to 2005 dollars using the
Chemical Plant Engineering Index. These costs are too old and they do
not reflect the large price increases in environmental controls that
have occurred during the last few years. Since ISIS is going to be used
for projecting policy impacts in the future years, the proper way would
have been to estimate the costs based on current prices and then adjust
them back to the year 2005 using the above index.

SKA.7:…This document has established average effectiveness levels for
various environmental controls for application on cement kilns. Some of
these levels are different (usually higher) than those presented in
another EPA document: “Assessment of Control Options for BART-Eligible
Source by NESCAUM, March 2005.” Can this be justified?

SKA.8: Page 16, Gas Treatment: Injection of biosolids, a potential NOx
control option, has not been addressed.

SKA.9: Page 26, Limestone Variable Costs: The cost of $15/ton is for
limestone, which is not pure CaCO3. Since all other data quoted in this
section is for pure CaCO3, the cost of limestone use would be higher
than $25.5/ton of SO2 removed, taking into consideration the impurities
in limestone.

SKA.10: Page 28, DLI: The reagent for this system would normally be
hydrated lime (which is what is quoted in the example in the second-last
paragraph). The cost of $75/ton appears to be for pebble lime, not
hydrated lime.

SKA.11: Page 33, Table 9: Suggest that you resolve the following
concerns that have been taken from other analyses on kilns, especially
the above BART-related document:

SCR and SNCR may not work on all wet and long dry kilns.

Dry lime injection may not work on all wet and long dry kilns.

In many cases only a tail-end type SCR system, with requirement for
reheating of flue gas, may be required. This may result not only in
additional costs, but also additional air emissions, if fuel is used for
the reheating.

Spray dryer technology may also be used for kilns other than wet type
kiln. Depending on site specific factors, this may be a more viable
option.

SKA.12: Page 34, Process Modifications: It appears that the performance
improvement options that may help reduce CO2 emissions have not been
fully developed. Options that may improve performance of raw material
preparation systems and clinker grinding systems have not been mentioned
at all. While conversions from one type of kiln to another, more
efficient type have been covered, they have not been included in the
input file. Such conversions may not be viable at all sites. However,
such limitations are not addressed. Use of high efficiency motors is
included in the input file. However, such motors may already exist in
some (or parts of some) newer kilns. It appears that a study to identify
a reasonable level of performance improvement that would apply to most
kilns on an average basis would be highly beneficial. 

Reponses to Comments on Appendix B

All comments to be addressed in the updated ISIS model/documentation.

HOB.1: Page 3-4: I would note that the USGS data pertaining to prices
for crushed stone ($/t) are for sales by crushed stone quarries. Thus,
they would be broadly applicable to prices for some of the raw materials
that are purchased by the cement company, but likely exceed what a
cement company “pays” for its own limestone. I would think that the
basic limestone cost (quarry, primary crushing, convey to raw mill)
would be typically ≤ $5/st.

HOB.2: The average hourly wage table look a bit low for a cement plant,
but I am not sure. Need to get some actual cement plant data. My main
issue is that the typical cement plant worker has 10+ years of
experience, I would think, and so would likely draw a higher wage.

HOB.3: Table 6: The electricity data seem high. First, whereas the logic
of the PCA’s “equivalent tons” may be sound for heat, it makes
less sence as the denominator for cement, given that roughly 1/3 of the
consumption is in the finish mill. The USGS data relate the unit
consumption to total cement (not just portland) production. We show for
2005: wet plants = 135.4 kWh/mt; dry plants (all types) = 139.0 kWh/mt;
combination plants (using wet & dry kilns) = 153 kWh/mt; and all
integrated plants combined = 138.8 kWh/mt. Grinding plants = 84 kWh/mt.

HOB.4: Table 7: the unit electricity prices may not be valid for large
industrial customers (they may get a lower rate).

HOB.5: Table 10: a few comments (reflecting newer info):

The Dixon, IL and Charlevoix, MI plants are now owned by St. Marys
Cement.

Hanson plant, CA is now owned by Lehigh SW

RMC plant in CA is now owned by CEMEX

Lehigh Allentown, PA plant is actually their Evansville plant

Holcim (US) now controls all St. Lawrence facilities in the USA.

Essroc’s Nazareth III plant in PA shut its kilns in 2005 (only kilns
are at Naz. I)

Lafarge’s Atlanta, GA plant shut its kiln in 2004; now only grinds
some clinker to make masonry cement, regrinds Type I into Type III
portland cement, and grinds granulated slag.

Dragon Cement was a dry plant in 2005 (the wet to dry conversion was
completed 4/04).

Reponses to Editorial Comments

DBED1: The narrative in the documentation is inconsistently
written.…there appears to be somewhat loose or inconsistent
translation of information in memos from Andover Technology Partners
into Chapter 2 and elsewhere. Some examples are mentioned below. 

…inconsistent inclusion of units of measure in the narrative

Conversion table to be added in the updated documentation.

…Table 2-2, 2-4 (SPC), and other tables do not have units labeled
clearly. 

Corrected in the updated documentation

Table 2-7 has lb CO2/MMBTU for CO2 but lbs/MMBTU for H20. Why not label
them symmetrically?

Corrected in the updated documentation

It also is difficult to track the units through equations in Chapter 3 

Chapter has been revised in the updated documentation.

DBED2: Page ix, middle. “…requirement compared to…” Does this
mean “in lieu of…”?

Corrected in the updated documentation

DBED3: Page 1-5, top. “…raw material feed…” This might be easier
to read if it said “feedstock.” 

Corrected in the updated documentation

HOED1: The text needs some editing, with especial attention to the use
of commas (you need more commas) and the common omission of the
indefinite article “the”. The word “data” is always plural: thus
“data are/were/show….”, never “data is/was/shows…”. There is
no need to capitalize portland cement (c.f., Portland cement), although
it is certainly OK to do so. The word “since” has a time context and
should be avoided where the meaning really is “because”.

Portland cement left as “Portland cement” in the updated
documentation.

HOED2: It is the U.S. Geological Survey, not U.S. Geological Service.

Corrected in the updated documentation.

HOED3: Use degrees Celsius (T°C), not °F.

Convention for the updated documentation is ___°F (___°C)

HOED4: All units, in all tables, must be defined. “Ton” does not
imply short ton. All (certainly most) tables need to be checked to see
that they include citations for the data source(s).

Done in the updated in the documentation

HOED5: As much as possible, avoid the use of acronyms in the text.

Acronyms were kept to the minimum in the updated documentation. 

HOED6: As much as possible, recast the equations into standard
mathematical notation rather than “computer printout” notation. For
example: IPG * CAPACITY ^0.4 is not as clear as it should be. Probably
CAPACITY was defined near the equation, as was IPG, and * is fairly
universally understood. But ^0.4 is not: is it the CAPACITY0.4 or is it
CAPACITY x 100.4? At least define it somewhere.

Chapter 3 has been revised in the updated documentation.

HOED7: Virtually everywhere, you use SO2, where SOx would probably be
the better term. But you could state that, for computational purposes,
you have used SO2 because it is “rigid” stoichiometrically and, in
any case, makes up the majority of the SOX. Ditto for CO2 including CO.

Comment addressed in the updated documentation.

HOED8: Avoid using MM for million, just use M. I think this is an issue
only with your Btu discussions. 

Explained this the first time MMBtu was used.

HOED9: The summary…“pyroprocessing,” it is generally more
applicable to the combination of calcination and sintering (clinkering),
and not so much to preheating and drying. 

Terminology has been updated in the updated documentation.

HOED10: You talk about “criteria pollutants” and HAP almost as if
they are interchangeable. It reads as if NOx and SOx are HAP. 

Criteria pollutants and HAPs are separate in the documentation.

HOED11: Para 1: say calcium aluminate hydrate, not calcium aluminum
hydrate.

Correction done in the updated documentation.

HOED12: if at all possible, avoid using PC for “precalciner” as it
is a common acronym for portland cement.

Correction done in the updated documentation.

HOED13: **Section 1.1.1**

“….rotary kiln…cylindrical furnace…”—I presume that the
latter is an explanation of the former, so put it in parentheses. The
material temperature is generally taken to 1450°C, not 1500°C (white
cement kilns probably do bring the material to 1500°C).

Done

“…the heart of the process generally is the rotary kiln….”

Correction done in the updated documentation.

The low end of the length range would more typically be c. 150 ft for a
precalciner kiln of c. 0.7 Mt/yr capacity.

Correction done in the updated documentation.

You state that the kiln is fired at the lower end, but you might wish to
also mention midkiln firing.

Correction done in the updated documentation.

The term “clinkerization” is not much used; more commonly, you see
“clinkering” or “sintering”. (Also “burning” but this is to
be avoided as too ambiguous).

Correction done in the updated documentation.

HOED14: Figure 1-2: in the caption, you might want to state that the
schematic for a precalciner kiln would be essentially identical.

Correction done in the updated documentation.

HOED15: Page 1-5: last paragraph: make sure it is understood that
“smaller” means lower capacity, not smaller dimensions on the kiln
tube.

Correction done in the updated documentation.

HOED16: Table 1-3: The data in this table represent U.S. Census trade
data (as supplied to and then compiled by the USGS). The original data
were in metric tons. DO NOT CONVERT THESE TO SHORT TONS!!!

Correction done in the updated documentation.

The percentage column needs clarification: percentage of what (USA total
imports).

Correction done in the updated documentation.

HOED17: Paragraph 1: a bit clumsy explaining where S comes from—it is
from sulfur-bearing compounds in both the fuels and the raw materials.
Many organic compounds (in fuels) contain sulfur, as do kerogens in
sedimentary rocks like limestone. Both fuels (especially coal) and raw
materials can contain sulfide minerals (chiefly pyrite or marcasite);
and sedimentary rocks may contain sulfates like gypsum.

Modified in the updated documentation

HOED18: Figure 1-6 is small to the point of uselessness. Why not make it
into 3 Figures (1-6a, 1-6b, 1-6c)?

Addressed in the updated documentation.

HOED19: Table 2-5: define “mg” (milligram or microgram?)

No “mg” found in table 2-5.

HOED20: The average CO2 emissions per ton of fuel noted in this section
should be rounded!…

CO2 emissions left as presented.

HYED1: Page 1-3 and 1-4. Figures 1-1 and 1-2 Update cement diagram –
add fabric filters (for PM controls) instead of ESPs and simplify the
diagram with just the cement making process

Figures replaced in the updated documentation.

HYED2: Page 2-3 and 2-4. Tables 2-2 and 2-3 specify the units

Units are being specified in the updated documentation.

HYED3: Page 2-8. Table 2-6 Change red color to be light green.

Table has been modified in the updated documentation.

Attachment A

Invitation Letter to Reviewers

Attachment B

Review Comments from

Dallas Burtraw, Resources for the Future

This review focuses on the operations of the ISIS model with a specific
example applied to the cement industry. The model makes available to the
regulatory process an integrated framework for policy evaluation of
environmental performance. The accomplishments in this model are
substantial. To be constructive and useful, however, I focus my comments
on points where there appears to be room for improvement. My comments
focus on three areas. One is the exposition of the documentation
narrative. This is important in order for the model to be widely
understood and to realize its full potential impact. A second is the
logic in the underlying algorithm and potential issues in the banking
function of the emissions module and in the investment algorithm. Third,
in some cases the algorithm seems to limit the applicability of the
model with respect to what may emerge as important policy questions in
the next few years. Data constraints may not require these limits in the
model. The two policy applications I mention specifically are episodic
NOx controls and potential climate policies that could not be addressed
without some modifications to the algorithm.

Before mentioning specifics, let me summarize further by saying that I
am thrilled that the modeling effort has been undertaken. Without
knowing the resources that have been committed one cannot say the effort
is cost effective; however, from my research perspective and judging by
the product, the effort to develop a model that can be used for policy
and regulatory analysis has been successful. Let me focus on what might
be opportunities for improvement. 

The narrative in the documentation is inconsistently written. Chapter 1
is very accessible. Chapter 3 is more technical, but nonetheless has
substantial opportunities for improvement. I give suggestions below for
reorganization. Moreover, there appears to be somewhat loose or
inconsistent translation of information in memos from Andover Technology
Partners into Chapter 2 and elsewhere. Some examples are mentioned
below. 

I am concerned that accessibility is harmed importantly by the
inconsistent inclusion of units of measure in the narrative. Adding
units or examples is the easiest way to help the reader understand the
model. In many cases I was unsure about units. Table 2-2, 2-4 (SPC), and
other tables do not have units labeled clearly. However, note that the
corresponding table in the documentation from Andover Technology
Partners (Sept. 23 2008) has units labeled. These should be copied into
the ISIS documentation. Table 2-7 has lb CO2/MMBTU for CO2 but lbs/MMBTU
for H20. Why not label them symmetrically?

It also is difficult to track the units through equations in Chapter 3
and the introduction of new data elements and variables seems to have
units are inconsistently described. It might be helpful if the labeling
of variables consistently used a label such as ‘F’ for a factor when
a variable is unit free (such as percentages).

The emissions policies that can be accommodated in the model represent
important options facing policy makers. However, evidence from other
research indicates that air pollution problems associated with
conventional air pollutants have changing characteristics as a
consequence of a changing climate. Ozone may become increasingly an
episodic problem with “peakier” peak ozone concentrations.
Similarly, longer periods of stagnant air may have regional
characteristics. A potentially valuable improvement in the model
capability would be to allow for emissions trading programs (or
regulatory standards) that have a location and/or time dimension. For
example episodic controls on NOx emissions may be an element of ozone
policy in the future. Episodic controls could be in effect for a period
of days, or they could apply during early morning hours of a given day.
They could be implemented within an incentive-based (trading or tax)
approach to policy through the trigger of a different permit or tax
level for a period of time. Could the model be used to simulate this
type of regulatory issue by incorporating information about the cost of
adjusting production schedules or fuels? Also, could the trading program
represented in the model have a regional dimension, such that different
emissions allowance prices are in effect in different regions and
seasons of the year? The memo from Andover Technology Partners (Sept.
23, 2008), page 5 bottom, indicates that calculations “apply for any
period of time so long that you are consistent with the time…” One
might infer that episodic operation of the facility would lead to
potentially important differences in its operating characteristics. Can
this be addressed in the narrative now and in a future version of the
model? Care could be taken now in the narrative so that the model is not
used inappropriately or its results misunderstood.

An important dimension of environmental impact of a facility is its
stack characteristics. These are not mentioned anywhere. This is
important if the integrated production cost ISIS model is to be combined
with a broader environmental integrated assessment. What is the stack
height? Do emissions of NOx from a kiln have similar parameters
(velocity, elevation, temperature) as emissions from a coal fired power
plant? Users of the model for integrated assessment will need to know
this, and how it varies across plants and regions.

In the policy experiment that is illustrated, the emissions bank expires
before the end of the second period as a consequence of the policy that
is modeled. Can the model accommodate a bank that does not expire before
the end of the second period, or is a terminal year date necessary for
the banking algorithm to converge? How would the terminal year be
determined? Is there a backstop technology that determines long-run
abatement cost?

An important additional dimension to emissions policy is the way that
allowances are allocated. In a competitive market it should make no
difference to product price (absent capital/liquidity constraints)
whether they are grandfathered or auctioned, as long as there is no free
allocation for new sources and, under grandfathering, there is no
surrender of allocations for changes in utilization or retirement of
existing facilities. However an important emerging policy issue is the
threat of unfair international competition under climate policy, and a
proposed remedy (including recent proposed legislation from Rep. Inslee)
is output based, updating, free allocation of allowances to industries
exposed to unfair competition. This type of allocation provides an
incentive to keep production onshore. Since this is an emerging issue it
would be a great enhancement to have it considered and modeled carefully
in ISIS.

In the power sector, an important part of NOx emissions reductions has
occurred through NOx “trimming” meaning that small changes in
combustion temperature are used to reduce the formation of heat related
NOx. This leads to slightly less efficient combustion and higher CO2 per
unit of electricity generated. I am confused as to whether it is a
factor in cement. The background technical memos discuss it somewhat,
but it does not appear in the model documentation.

P. 1-11, Figure 1-7. Here and in the narrative the reader might be
initially surprised to see “Emissions” as an input to the ISIS
engine. Emissions would seem to be an output of technology and
production choices, and indeed they appear in the “Model Outputs”
node. I understand, and it is explained in a subsequent chapter, that
the emissions inputs are describing baseline or uncontrolled emissions
at baseline levels of production. This should be explained in the
context of Figure 1-7. For example, at some facilities there already are
low-NOx burners, and at others there are not. Technologies introduce an
emissions modification factor that leads to ultimate emission estimates.

What constraints affect the choice of technologies for pollution
reduction? Can only one technology be chosen? Can it be changed? That
is, can a technology chosen in year 5 and then removed and replaced in
year 10 with another technology choice? This appears to be happening in
Figure 5-4 on page 5-4. See, for example, the rise and fall of
LNB+CEMStar. When this can happen, how are capital costs considered? Are
opportunity costs properly accounting for only marginal going forward
costs (avoidable costs)? 

It is not clear to me how costs are considered in the model. Are
investment decisions using more than an annualized capital cost to
consider in-place controls? It must be true that the expected value of
the stream of future revenues and costs justify an investment; not just
the revenues and costs in one year, considered one-year-at-a-time. It
should be that investments cannot be reversed, and the calculation of
net benefits includes the present discounted value over all options.

One could move the Objective function in Section 3.6 up to the beginning
of Chapter 3. I would create a table, so that on the right hand column
each term definition is provided as one works one’s way down the
equation. That is, put definitions following “where” in the right
hand column, spaced appropriately. Then one can “read” this
objective function and see the ways that economic intuition is satisfied
in the formulation of the objective function. Then, the specifics of the
algorithm can be explained in the other parts of Chapter 3.

Are transport-related emissions included in the emissions calculations?
This concerns transport of fuel and material to the plant, and transport
of cement to market. Presumably not, but this could be indicated. 

In the narrative it wasn’t obvious to me whether and how
transportation costs are included in the model. Are the costs of all
materials and fuels the delivered costs? Secondly, it is not described
how are costs of transporting product between regions accounted for in
the narrative, although cost differences are described. From the Excel
spreadsheet input data files it is clear there is a region-to-region
transportation dataset. This should be explained in the model
documentation. The optimization problem introduces the new term
“transcostt.” Regarding this and other new terms, it would be
helpful if these terms are not explicitly introduced earlier, at least
to indicate earlier that the related considerations are included in the
model explicitly.

Location in a region should matter. What assumption can be made or is
made about the location of the new unit and the expanded unit in each
region? Are all input costs uniform for all plants within a region? 

Further, what can be said about the location of new capital investment
is relevant to subsequent modeling of emissions impacts because latitude
and longitude information is required for the next step in an integrated
environmental assessment. The model user will have to have information
about which plant within a region expands, or where new plants are
located. A reasonable assumption may be to distribute new capacity
proportionately at the location of existing facilities.
Probabilistically this may represent the future. Whatever may be
reasonable, the model narrative could provide some guidance for the user
about what assumption is recommended.

It is clear from the optimization problem that emissions policy
compliance costs (taxes, permits) are included in the variable cost
calculation. But I did not see this stated explicitly anywhere in the
narrative. It should be mentioned in chapter 1.

CEMStar is referred to in passing in various places. Is footnote (a) in
Table 2-8 the description? It is described in Andover Technology
Partners (9/2508) page 34. Some short summary should be included in the
ISIS model documentation.

Accounting of capital cost is not transparent to me. I see the
market-clearing price is determined by the variable cost of the marginal
capacity on an annual basis. The marginal capacity in year 10 may be the
newly added capacity (e.g. capacity built in year 5). It is not apparent
whether the variable cost in year 10 includes capital cost for the newly
built capacity (long run marginal cost). The proper choice depends on
the measure of competitiveness in the industry and the role of business
cycles. In a downturn, say in year 10, the recent capital investment is
sunk, and production should depend only on year 10 variable cost.
However, in the absence of business cycles one might expect to see long
run marginal cost as a proxy for the expected (or average) marginal cost
in future years. See for example the discussion on page 3-2: “Note
that capacity is added endogenously…” The determination of variable
cost and product price is critical to this endogenous calculation. See
also page 3-4, the description of annual fixed cost. Perhaps the more
appropriate question is whether annual fixed costs is included in the
calculation of market prices. Is annual fixed cost considered a variable
cost when the model is solving in annual time steps? The narrative here
could be strengthened.

Calibration is a standard operation to account for aspects of market
structure and performance that are unobservable in the input data. It
would be helpful in Chapter 4 if an explicit equation showed the
variable cost calculation with the calibrator constant included so that
one could see the level of detail in the calibration. Also, see Table
4-1. The “Total” in the table is labeled as a % Change of 5.1 but
the numbers indicate a change much less in the total.

P. 4-4, middle. “A single calibration constant…” Why is a single
calibration constant used to adjust variable cost of production? The
justification for calibration may be market power, which is likely to be
regionally specific. The assumption of the model is competition. If
there is not a perfectly competitive market, or if there are regional
rents associated with transportation costs, presumably these would be
captured by the calibration process. It appears calibration does not
occur regionally or seasonally, which may be an important limitation of
the model. 

What is the plan for updating the data inputs and technology
characteristics? The vintage of these inputs should be indicated. For
example, P. 1-7, top. The projections for capacity expansion are likely
out of date now, due to financial market issues. If this projection
cannot be updated, at least a note should be added to indicate that
these projections predate economic events of late 2008.

On page 2-7, bottom paragraph, I had a hard time understanding the
algorithm described. I interpreted this to say that pyroprocessing
captures sulfates, but sulfates are usually not considered a gas, and
the sentence describes gas passing through the calcinations zone. (In
the memo from Andover Technology Partners (September 23, 2008) p. 14, it
is stated “SO2 released from feed materials…” It does not mention
sulfates in this instance.) Further, preheater and precalciner kilns are
more effective…“Accordingly,…” Does this mean that in these
types of kilns or in the dry and wet kilns? Also, location “appears to
have a significant role in SO2 emissions…” Does the feedstock in
reference not include fuel? 

P.2-23. The elasticity value cited is -0.88. Is this to be interpreted
as a short run or long run elasticity? Is there reason to think they
differ, or not? Are there long run substitutes for cement?

P. 3-2. It is crucial to indicate when the constraints limiting capacity
changes actually bind when reporting model results. What is the
relationship between the chosen values and historical experience? I am
concerned about these constraints because if they bind under a strict
climate policy then one might ask if the inclusion of the constraints
and the role of these constraints should be considered in sensitivity
analysis. If the constraints are binding and this is not apparent to the
model user and policy analyst it could lead to an inappropriate
prediction of what might happen under strict climate policy. 

Similarly, on P. 3-3 we learn that imports are constrained to historic
experience. However, many are concerned that under strict domestic
climate policy there will be infusions of imports from countries that
have not take on climate-related obligations. ISIS could not be used to
inform this question. At least, this constraint should be tagged and
output should be clear when, if ever, it binds.

Minor comments:

P. ix, middle. “…requirement compared to…” Does this mean “in
lieu of…”?

P. 1-5, top. “…raw material feed…” This might be easier to read
if it said “feedstock.” 

P. 1-5, bottom. How does one interpret “kiln capacity?” Is output
measured per unit time? I learn in the memo from Andover Technology
Partners (September 23, 2008) in a footnote on page 1 that capacity is
normally expressed “per hour.” This should be added throughout the
ISIS documentation.

Also, the “Production Technologies…” node could have more
information in the box. Is technology choice a decision made in this
node?

Page 1-9, bottom. Local constraints such as nonattainment areas and PSD
areas are described, but there is no indication whether these are
represented in the model or have been ignored.

Page 2-2. “The ISIS model determines the optimal fuel type…” Does
the model account for long run contracts for fuel, or are all projected
fuel choices assuming a spot market for fuel?

P. 2-5. “In ISIS, projected <delete: kiln> retirements of existing
kilns…”

P. 2-7, Table 2-5. Explain sources in this table and why they diverge.

P. 2-7, bottom. “…location (i.e. feedstock) appears…” Does this
mean to say: “location and the associated feedstock (fuel type)
appears…”?

P. 2-8, Table 2-6. The colors do not appear consistent. For example, why
does Birmingham, with SO2 emissions of 0.09 for the PC technology, have
a white color in the cell?

P. 2-10. “By-product benefits” Also costs?

P. 2-10, Table 2-8. Low NOx burners are usually thought to reduce
efficiency in coal-fired electricity generation. Is there no penalty in
a kiln? I expect this would appear under CO2. However, the memo from
Andover Technology Partners (9/25/08) page 5 indicates that there is a
penalty of 0.09 kW per short ton, which does not appear in Table 2-8.
Also, there is no CO2 penalty identified with the use of a Wet Scrubber,
but later on page 3-5, a penalty associated with “generation of CO2 in
a wet SO2 scrubber” is described.

Table 2-8 also seems inconsistent with the memo from Andover Technology
Partners with respect to tires and the CO2 implications. The memo
clarifies that the use of tires must be compared against the fossil fuel
that is being replaced.

P. 2-12, Table 2-10. It may be useful to add an escalation factor for
construction materials. This is to be newly added to the AEO, I
understand.

P. 3-2, middle. I don’t see how the constraints relate. It appears
that 2% can be replaced in a given year and no more than 10% over a
3-year period. But no more than 6% could be replaced in a 3-year period,
based on the first constraint.

Bottom. “Note that replacements are associated with individual
units….expansion…changes are…in regional capacity.” Then why are
they summed?

P. 3-5. What is “priceration” in equation 3.2.5?

P. 3-5, middle. Energydisl refers to secondary fuel and polbasesubfuel
refers to substitute fuel. Are these the same thing? If not, what is the
difference? Elsewhere the term “fuel displacement” is used. These
are awkward labels. It would be simpler to refer to fuel substitution.

Equation 3.3.2. What is “cp”?

P. 3-7. I finally lost the thread in equations 3.3.7 to 3.3.9. Can some
narrative be added? Why is the sum in 3.3.9 divided by the number of
pollutants? What are units throughout these equations?

P. 3-9. An example is “catconsumptcost,” which is not defined as a
price times a quantity. However, see 3.4.11a. There, cost is defined as
a price times a quantity. This inconsistency makes it unclear when
rates, percentages or actual units are relevant to various equations. 

P. 4-4, top. “the marginal <?> of the regional…”

Attachment C

Review Comments from

Kevin Culligan, US Environmental Protection Agency

To:		Wojciech Jozewicz, ARCADIS

From:		Scott Benolkin, Clean Air Markets Division, EPA

Date:		9 February 2009

Subject:	ISIS Cement Comments

ISIS Cement is a thoughtful, well-executed effort to directly address
sector compliance issues that other models assume away. It will fill a
gap in EPA’s modeling and meet an increasingly important need for
detailed analysis of the environmental and economic impacts of air
policy. As the first in a planned family of models, ISIS Cement also
serves as prototype for a more general concept, and therefore, its
structure and implementation bear more significance than those of a
one-off model. With this in mind, I offer the following comments and
questions:

The documentation (2.2.3.2) mentions a cement demand elasticity of
-0.88, yet the model results show no changes in demand in the policy
case, despite significant allowance prices. Is this a disabled feature?
Certainly, demand response will be an important factor in air policy
impacts, especially for greenhouse gases.

Lacking control technologies for CO2, ISIS Cement is not yet ready to
address climate policy except as it relates to electricity and import
prices. For the former, electricity use should be a more prominent
output. For the latter, given the cement industry’s energy intensity
and the role of imports in meeting cement demand even without greenhouse
gas policy, the richness of import-price-related modeling will be
critically important to assessing the cement industry’s response to,
for example, a price on carbon emissions. Both CO2 mitigation technology
and product substitution in demand response are sine qua nons for
modeling the cement industry’s response to climate legislation.

The rationale behind the choice of region definitions needs more
documentation, especially because the regions differ in some cases from
those in the EPA (1998) RIA cited (e.g., see Attachment 8 of that
document). Given the location of many plants far from their regional
market’s center (see Figure 2-1), assuming uniform regionwide
parameters could systematically underestimate the potential competition
and exchange between neighboring plants in different regions. As a check
of the assumption, it would be nice to know or estimate how much of each
plant’s commerce is within its ISIS region versus outside of it. This
is likely to be a significant assumption for other ISIS industry models
as well.

Uncontrolled NOX emissions intensities vary not only among but also
between average estimates by a factor of 2, ranging from the EPA (2000)
estimates shown in Table 2-5 to those Andover Technology Partners
provides. Have you examined the sensitivity of results to these
parameters? More practically, how much investment in monitoring
infrastructure would kilns currently need to comply with either
cap-and-trade or rate-based policies?

OMB’s 7% discount rate makes sense for analyzing the costs and
benefits of policy results from ISIS, but it is not a basis for
reflecting the discount rate cement companies use to make decisions.
IPM, for example, derives its capital charge rate from assessments of
actual discount rates faced in financing projects.

Equation 3.2.1 appears to set annual fixed costs proportional to a
unit’s production rather than its capacity. If I am interpreting this
correctly, it is a serious conceptual mistake.

Estimating and predicting based on a range of years rather than just one
for each would improve the calibration procedure. What makes ±10% a
reasonable range for predicting values one year later—i.e., how does
it compare to the typical annual variances of production, imports,
emissions, and prices? It seems like a much easier target for some
numbers than for others.

Variable cost seems like an important parameter to be imputing a
constant adjustment from a regression over (it appears) two years of
data. How sensitive are the results to the choice of this constant (or
the choice of using a constant)? [Note: my own limited investigations
tweaking that parameter in Inputs_Imports.gms show mostly level shifts.]

Constant escalation factors for fuel prices seem too simplistic and,
negative factors contradict recent projections (see comment 10).
Ideally, a supply curve informs the cost of fuels. Absent that, ISIS
should use actual annual or multi-annual projections rather than an
annualized 20+-year average. IPM, for instance, can supply such
projections by region and coal type under a given policy.

Also, while the model’s energy prices escalation factors reconcile
with those in AEO 2008 Table A1 (with the exception of using minemouth
rather than delivered coal prices), they disagree with those in Table A3
(by sector), which for instance has coal rising, not falling, in price.
Notably, more recent projections by EIA and IPM have much higher future
price increases (2–3% annualized) than AEO 2008.

General questions:

How substitutable are the different varieties of Portland cement with
each other and with masonry cement? How well do prices and demands for
different varieties correlate? What about other substitutes for cement?
At the very least, knowledge of these options can inform more realistic
demand response (as in comment 2 above).

Given transportation costs and the fact that 10 companies account for
80% of production overall, is market power an issue in any regions or
subregions?

Are PCA estimates (e.g., of demand and capacity expansion) corroborated
by independent sources? More immediately, is the PCA still projecting 27
new kilns through 2010?

In the real world, do producers buy at a world market price or benefit
from multinational ownership?

Short of process changes like CEMStar, are marginal changes in
feedstocks available to kilns in the real world? How much difference
does this make?

Attachment D

Review Comments from

Fereidun Feizollahi, CARB Economic Studies Section

Overall, this tool could probably be put to some use, however, at the
moment I can not tell if the model is functioning correctly or if there
are problems with the output spreadsheet.  

With cap and trade, the results appear to be as expected in terms of
direction of change and the resulting price effects.  However, some of
the results (e.g., policy production and policy imports) did not seem to
be influenced by the required reductions, that is, I got the same
results for a small CO2 cap as I did for a sizable NOx and SO2 cap and a
small CO2 cap.  Some variables such as the average policy price did
change but not significantly given the difference in the scenario.  

Differences were seen when I allowed for banking.  With banking, policy
production and policy imports were affected, but only for years
2010-2016.  Years 2017-2025 were the same as the no banking results. 
The price difference between a banking scenario vs. a no banking
scenario seemed a bit excessive: far greater than that seen with a
significant change in the program scope (NOx, SO2 and CO2 vs. CO2 only).
 Also note than when banking is not allowed, the resulting allowance
price is reported as essentially zero (eps).

When comparing different fees, again production and imports seemed to be
the same regardless of the amount of the fee. 

Comments on EPA’s ISIS Model

Demand projections -   In ARB’s draft calculations for California’s
cement industry, ARB staff has assumed a 2% annual growth rate in cement
demand.  Due to the economic downturn, staff will be reevaluating the
annual growth rate.

Production and control costs and associated escalation rates - 	ARB has
limited information in this area and can not provide comments.

Treatment of imports - California cement facilities will be part of a
cap and trade system to reduce greenhouse gas (GHG) emissions.  Details
of the cap and trade system are under regulatory development.  In
addition, ARB staff is considering the development of a regulation to
establish blending requirements of supplementary cementitous materials
at concrete batch plants.  The treatment of imported cements will be
considered as part of that effort.

Attachment E

Review Comments from

Sikander Khan, US Department of Energy

Comments submitted by Sikander Khan included comments by Chris Nichols
and Charles J. Drummond

Comments by Chris Nichols and Charles J. Drummond

U.S. Department of Energy, National Energy Technology Laboratory

January 29, 2009

General Comments

A model of the domestic industrial sector that encompasses a
comprehensive approach to emission of pollutants for all three media
(air, water, and solids) would provide significantly greater value than
a more limited focus. All three media need to be represented to truly
understand pollution reduction approaches that minimize environmental
impact and avoid simply reducing air pollution by increasing water
pollution. Current descriptions of the ISIS model reference all three
media; however, this focus needs to be expanded. I recommend that,
consistent with NAS 2004 recommendations, a long-term vision for the
ISIS model be established with an overall objective to optimize air,
water, and solid releases of pollutants from industrial sectors. This is
not a trivial problem. However, the significant value of achieving this
vision motivates the effort that would be required. It is critical to
establish this high level vision now so that a model framework capable
of accomplishing the vision can be established. Efforts to build the
model can be planned on an incremental basis as long as the framework is
established at the beginning. It would be too difficult to adapt a model
developed primarily to optimize reductions of air emissions to conduct
optimization of all three media if the concept for this capability was
not included in the initial design of the framework.

An ability to store multiple scenario cases would be good. For example,
if you ran two policy cases scenarios back to back (e.g., a CO2 cap &
trade vs. tax), it would be useful to be able to compare the two easily.

How are the regions modeled for cement going to map into other
industries (especially NERC regions for electricity generation)?
Integration with other industries is going to be really vital for
getting valid results in any cap & trade scenarios in this model. This
would be especially important if costs are higher for one industry than
another to control certain pollutants.

The model seems to rely heavily on imports in very restrictive scenarios
in order to meet emissions requirements. I could not find more detailed
information on how imports are treated – is there a cost curve that
increases as more imported cement is demanded? Also, the documentation
states that “selected base year (2005) regional import information was
used to estimate import levels in future years.” How? Does it
proportion future imports the same way they were divided in 2005?

Interregional trade seems like it would be very sensitive to fuel prices
– using diesel prices avg. 1994-2005 may be misleadingly low.

Are there any ways to add CO2 capture options? This is going to be
really important for future integration, I think. Also, when modeling
CO2 scenarios, incorporation of domestic and international offsets are
going to be really important to include.

Specific Comments - ISIS Overview Presentation (1/14/2009)

Slide 6 – I agree that the objective should be to minimize costs of
both production and emission control subject to meeting both demand for
the commodity and emissions requirements. All of these criteria are
important for a realistic model.

Slide 7 – Inclusion of both emissions controls and modification of
industrial production equipment as choices that the model can make to
reduce emissions is necessary and valuable.

Slide 14 – Starting and ending conditions for banking of emissions
credits are not realistic. The model needs to establish constraints on
banking that make the emissions banking process sustainable beyond the
forecast horizon. It is unrealistic (and overly simplistic to the point
of being meaningless) that all banked credits are consumed by the end of
the forecast period. This would only occur in reality if the end of the
forecast period coincided with the end of the emissions reduction
requirement.

Specific Comments - ISIS Model Documentation (12/23/2008)

Page 2-7; Section 2.2.1.8 Emissions: This section only discusses air
emissions. A discussion of all substances emitted to all media would
provide greater value.

Page 5-1; Section 5.1.1: The emissions cap and trade example provided is
overly simplistic. It does not provide insight into the application of
the ISIS model to actual policy analysis. In the example given, all
banked allowances are consumed by the end of the second phase of the
rule. In actual application of the model, banking of allowances would
most likely continue beyond the end of the forecast horizon being
modeled. The capabilities of ISIS to model a realistic scenario should
be provided in an example. Constraints on conditions at the end of the
forecast period would be needed to maintain realistic quantities of
banked allowances that would sustain industry’s approach to meeting
the rule beyond the end of the forecast horizon.

Specific Comments - ISIS Model Output Spreadsheet

AggregateResults tab:

Cell F6 should have some sort of label telling what it is.

Fuel savings (cell J4) was not able to generate any scenarios where that
was populated - what is it?

Rows 9-24 would be more helpful if the formatting had commas, truncated
decimals to one or two places.

What does “eps” mean? That occurs frequently throughout the output
spreadsheet.

I ran a CO2 reduction case (50% 2010-2014, 70% 2015-2025) but no
allowance price was calculated for CO2 in the Output spreadsheet – I
think this would be a useful output.

Cells AL30:45 – these are not labeled, and it’s not clear what this
is.

Rows 53-68 and 72-87 what are the units? Any significance for the red
font?

In the CO2 case I ran above, in 2025, ISIS expanded capacity by 23.5 and
built 5.3 in new capacity and seemed to keep all existing capacity,
compared to 22.4 and 3.8 respectively in the base case. This is
counter-intuitive to me since overall domestic production is less in the
policy case – why would the model build new capacity and not use it if
it is not retiring older dirtier capacity?

Specific Comments - ISIS Model Input Spreadsheet

TransCosts tab:

Cell C2 – do you mean Newark, DE?

Would be helpful to be able to adjust diesel prices – using AEO2005
may not be the most reliable data.

Emissions tab:

Looking at the documentation, SO2 and NOx have lots of data on their
emissions factors, but CO2 does not. The emissions factors for CO2 (rows
36-38) seem lower than those in EIA’s Table 6-1, Documentation for
Emissions of Greenhouse Gases in the U.S. 2006.

Policy tab:

For the % reduction policy options inputs (cells F5:H22), what if the %
reduction compared to? Is it a year-by-year? So when 60% is entered for
2015, is it 60% of the baseline’s 2015 emissions, or 60% of the
starting year’s emissions?

Another useful input would be to be able to manually enter an emissions
cap, i.e., enter the maximum emissions level for each year.

K5:M5: don’t understand “miner” in these cells.

Additional Comments by U. S. Department of Energy

February 2, 2009

Comments on Industrial Sector Integrated Solutions Model, December 23,
2008:

Page 1-5, Table 1-1: Use of two sets of numbers for kiln specific heat
input (or energy intensity) in this table and Table 2.4 is confusing. An
explanation of the difference would be helpful.

Page 2-2, First Paragraph: It is stated that the ISIS model determines
the optimal fuel type. What is the basis of this selection? Is there any
information available for the existing fuel type, e.g., for the base
case? Can the fuel type be an input to the model?

Page 2-7, Section 2.2.1.8: In the first paragraph, it is stated that
ISIS model uses emission intensities to estimate NOx, SO2, and CO2
emission projections. However, from the model input file (Spreadsheets
“Units” and “Emissions”); it is not obvious if this has been
done with the CO2 emissions. It appears that CO2 emissions for the year
2005 are calculated by multiplying the clinker yearly production rate
with the same factor (0.45 for fuel) for all kilns. Also, the 2005
baseline figures for NOx and SO2 emissions are estimated based on a
constant multiplier for all kilns. Notes 10, 11, 12, and 14 are not
provided, which may have explained the reasons.

 Page 2-9, Table 2-7: This table lists different CO2 emission rates for
PRB and bituminous coals, as estimated by J. Staudt of Andover
Technology Partners in Appendix A. However, in the model input file,
only one type of coal (bituminous) is used. It is assumed that the model
has the capability to use different coal types, if input data were so
prepared.

Page 2-11, Section 2.2.3.2: The economic parameters listed do not fully
report the calculations built into the model for the overall economic
analysis. For example, what is the basis for these calculations: current
or constant dollars? How are the construction times for the controls are
taken into account? Does the model estimate some of the construction
costs (e.g., allowance for funds used for construction), since such
costs may not be included in the data provided in the Appendices? In
Table 2-10, what does the annual escalation rate of 3.09% for the
variable O&M cover – all variable costs or some of them? If all, then
why different factors for limestone and gypsum? Is there a separate
factor for lime as well?

Chapter 3, ISIS Mathematical Framework: Can the model accommodate
different safety valves built into a policy framework. Can a case for
cap-and-trade with a limit on allowance price (with other required
caveats) be run?

Page 5-5, Figure 5-5: Can there be a scenario, where the price of cement
is affected to the point that it starts to affect demand? 

Page 5-6 Table 5-1: The increase in the combined import and control
costs ($2 billions) does not match the increase in the total cost ($1.58
billions) for the example case.

General Comment: The costs associated with the base and policy cases
should be reported in terms of factors (e.g., $/ton of clinker or $/ton
of air pollutant removed) that can easily be understood and compared to
the effects of other similar policies.

Appendix A, Andover Technology Memorandum, September 23, 2008:

Page 1, Table 1: It needs to be clarified if the specific fuel
consumption data provided includes only kiln fuel or non-kiln fuel
sources, such as raw material dryers, auxiliary boilers, etc. These
non-kiln sources can add significant CO2 emissions.

Page 4, Second Paragraph: The figures quoted in this paragraph are based
on pure limestone. It’s not clear if all of the figures in this
paragraph, e.g., for the steelmaking slag, are based on pure limestone.

Page 6, Total CO2: What is the assumption used in the model with regards
to the cement kiln dust, which is also a source of CO2 emission?

Page 7, Table 4: The range value (1.9-3.1) for the wet kiln NOx
emissions does not cover the average value (6.2).

Page 27, Appendix 4: The heating value of natural gas, 26,037 Btu/lb,
shown in the last column is too high. This value is generally between
21,000 and 22,000 Btu/lb. The reason for this may be the assumption that
there is only carbon and hydrogen in the gas. This is not true, since
there may be CO2, nitrogen, and other impurities present.

Appendix B, Andover Technology Memorandum, September 25, 2008:

General Comment: Many of the cost sources used in this document are from
years prior to 2005 and they have been adjusted to 2005 dollars using
the Chemical Plant Engineering Index. These costs are too old and they
do not reflect the large price increases in environmental controls that
have occurred during the last few years. Since ISIS is going to be used
for projecting policy impacts in the future years, the proper way would
have been to estimate the costs based on current prices and then adjust
them back to the year 2005 using the above index.

General Comment: This document has established average effectiveness
levels for various environmental controls for application on cement
kilns. Some of these levels are different (usually higher) than those
presented in another EPA document: “Assessment of Control Options for
BART-Eligible Source by NESCAUM, March 2005.” Can this be justified?

Page 16, Gas Treatment: Injection of biosolids, a potential NOx control
option, has not been addressed.

Page 26, Limestone Variable Costs: The cost of $15/ton is for limestone,
which is not pure CaCO3. Since all other data quoted in this section is
for pure CaCO3, the cost of limestone use would be higher than $25.5/ton
of SO2 removed, taking into consideration the impurities in limestone.

Page 28, DLI: The reagent for this system would normally be hydrated
lime (which is what is quoted in the example in the second-last
paragraph). The cost of $75/ton appears to be for pebble lime, not
hydrated lime.

Page 33, Table 9: Suggest that you resolve the following concerns that
have been taken from other analyses on kilns, especially the above
BART-related document:

SCR and SNCR may not work on all wet and long dry kilns.

Dry lime injection may not work on all wet and long dry kilns.

In many cases only a tail-end type SCR system, with requirement for
reheating of flue gas, may be required. This may result not only in
additional costs, but also additional air emissions, if fuel is used for
the reheating.

Spray dryer technology may also be used for kilns other than wet type
kiln. Depending on site specific factors, this may be a more viable
option.

Page 34, Process Modifications: It appears that the performance
improvement options that may help reduce CO2 emissions have not been
fully developed. Options that may improve performance of raw material
preparation systems and clinker grinding systems have not been mentioned
at all. While conversions from one type of kiln to another, more
efficient type have been covered, they have not been included in the
input file. Such conversions may not be viable at all sites. However,
such limitations are not addressed. Use of high efficiency motors is
included in the input file. However, such motors may already exist in
some (or parts of some) newer kilns. It appears that a study to identify
a reasonable level of performance improvement that would apply to most
kilns on an average basis would be highly beneficial. 

Is there any plan for showing the indirect effect on air emissions of
changes in the electricity consumption caused by modifications required
by policy decisions at cement plants? 

Attachment F

Review Comments from

Andy O'Hare, Portland Cement Association

EPA is in the process of receiving the peer review comments from the
Portland Cement Association.

Attachment G

Review Comments from

Hendrik van Oss, US Geological Survey

Hendrik G. van Oss, Cement Commodity Specialist

U.S. Geological Survey, Reston, VA 20192

January 29, 2009

I have broadly reviewed the ISIS document package that you sent to me on
a CD ROM (as well as the ancillary/replacement Excel tables). As I noted
to you earlier, I did not have time to run the actual software. I am not
trained in finance, so I have not looked at your costing algorithms in
detail.

Overall, the package is impressive—it represents a lot of work! I have
made a number of general and specific comments (see below), but,
overall, and assuming certain changes are made, I think that the project
will prove to be very useful in evaluating options/estimating the costs
of various (chiefly environmental-related) upgrades at plants.

General Comments

Editing

The text needs some editing, with especial attention to the use of
commas (you need more commas) and the common omission of the indefinite
article “the”.

The word “data” is always plural: thus “data are/were/show….”,
never “data is/was/shows…”.

There is no need to capitalize portland cement (c.f., Portland cement),
although it is certainly OK to do so.

The word “since” has a time context and should be avoided where the
meaning really is “because”.

Global

It is the U.S. Geological Survey, not U.S. Geological Service.

Very important: the U.S. Govt. has a mandate to use metric units as much
as possible. This report has taken data that are largely published as
metric tons (e.g., recent PCA data, USGS data, Census trade data) and
converted them to short tons. Then, in the text, you have always
provided a parenthetical conversion: “…9,217,487 tons (8,361,966
metric tons)…” –this is not only visually distracting (to the flow
of the text) but greatly adds to the length of the text. The use of
short tons in the tables has prevented me from conveniently checking
many of your data against my own (USGS) publications. I think that the
whole report (maybe not the Anderson addendum report) needs to be recast
in (only) metric units. One exception can be the use of Btus instead of
joules—as Btus are almost universally understood.

Use degrees Celsius (T°C), not °F.

All units, in all tables, must be defined. “Ton” does not imply
short ton. 

All (certainly most) tables need to be checked to see that they include
citations for the data source(s).

As much as possible, avoid the use of acronyms in the text.

As much as possible, recast the equations into standard mathematical
notation rather than “computer printout” notation. For example: IPG
* CAPACITY ^0.4 is not as clear as it should be. Probably CAPACITY was
defined near the equation, as was IPG, and * is fairly universally
understood. But ^0.4 is not: is it the CAPACITY0.4 or is it CAPACITY x
100.4? At least define it somewhere.

Virtually everywhere, you use SO2, where SOx would probably be the
better term. But you could state that, for computational purposes, you
have used SO2 because it is “rigid” stoichiometrically and, in any
case, makes up the majority of the SOX. Ditto for CO2 including CO.

In many places, you have used PCA data instead of the USGS data. In some
cases, this is fine—the PCA does its own survey of energy use & plant
capacity (and the USGS holds all plant-specific data proprietary). But
you may wish to point out to the reader (potentially familiar with both
data sets) that there are some important differences. On the kiln count,
for example, the USGS counts all kilns active for at least 1-day during
the year, whereas the PCA only counts the kilns that were active at
yearend (we count 183 kilns in 2005, the PCA just 181). Many of the PCA
energy data exclude white (portland) cement, whereas these are included
in the USGS data. The PCA provides the cement companies with heat
conversion factors for the various fuels, whereas the USGS asks the
industry to directly supply the Btus realized (at whatever HHV (gross
heat) conversion factor the plant normally uses). Any State-level (and
USA overall) consumption or production data published by the PCA are
taken from USGS data and you should cite the USGS for such data, not the
PCA. 

Avoid using MM for million, just use M. I think this is an issue only
with your Btu discussions. In my experience, M for thousand survives in
the world only in old Latin texts and for Copyrights, and in billing for
natural gas (MCF).

Somewhere, you should discuss the issue of data accuracy. In general,
the cement industry calculates, but does not weigh, its clinker and
cement production. They weigh their sales, however! In many places, data
ought to be rounded. Few data reported are good to better than +/- 1%.

Specific Comments

Summary

The summary is a little casually written. Although it is debatable as to
what constitutes “pyroprocessing,” it is generally more applicable
to the combination of calcination and sintering (clinkering), and not so
much to preheating and drying. 

You talk about “criteria pollutants” and HAP almost as if they are
interchangeable. It reads as if NOx and SOx are HAP. 

The Summary seems to lack discussion of CO2—you need to at least
mention it (and then you can say that your cost modeling was confined to
NOx and SOx. 	

Sec. 1.1

Introduce this with portland and blended cement (both are used in
concrete), and then state that you will focus on portland. But be
careful: the USGS monthly data (1/98 onwards) split out the two (Table
2a vs. 2b) and they should be added together for discussions of
consumption. I do not know if the PCA, in using our data, has combined
these in all cases. The USGS annual data includes blended cement within
the portland umbrella. The annual report (Mineral Yearbook) shows the
combined monthly sales (consumption) data in Table 9.

Para 1: say calcium aluminate hydrate, not calcium aluminum hydrate.

Sec. 1.1.1

“….rotary kiln…cylindrical furnace…”—I presume that the
latter is an explanation of the former, so put it in parentheses. The
material temperature is generally taken to 1450°C, not 1500°C (white
cement kilns probably do bring the material to 1500°C).

“…the heart of the process generally is the rotary kiln….” 

The low end of the length range would more typically be c. 150 ft for a
precalciner kiln of c. 0.7 Mt/yr capacity.

You state that the kiln is fired at the lower end, but you might wish to
also mention midkiln firing.

The term “clinkerization” is not much used; more commonly, you see
“clinkering” or “sintering”. (Also “burning” but this is to
be avoided as too ambiguous).

Very important (Global): it is certainly OK to distinguish wet vs. dry
kilns. But it is NOT acceptable to use “dry” to mean (only) a dry
kiln lacking a preheater and/or preheater/precalciner. You should call
this type of kiln a “long dry” kiln. The 3 general categories for
dry kilns are: 1) long dry kiln; 2) preheater kiln; 3)
preheater/precalciner kiln (which, after spelling out the first time,
you may shorten to precalciner kiln).

(Semiwet vs. semidry: frankly, these kilns can go either way. If the
kiln has a preheater and/or precalciner, the USGS lumps them within the
overall Dry kiln category. I am aware of only 1 plant of this type
(Buzzi’s Greencastle, IN plant).

Figure 1-2: in the caption, you might want to state that the schematic
for a precalciner kiln would be essentially identical.

Table 1.1: The data are typical averages, but are not universal. The
reader may think these are fixed data. The EPA source sited likely used
either USGS or PCA data.

Global: if at all possible, avoid using PC for “precalciner” as it
is a common acronym for portland cement.

p. 1-5: last para: make sure it is understood that “smaller” means
lower capacity, not smaller dimensions on the kiln tube.

I would note that the USGS Minerals Yearbook chapters on cement (MYB)
also allow the average capacity to be discerned: it would calculate as
0.577 Mt/yr (million metric tons per year) in 2005 (MYB, table 5).

It may be useful to discuss what is meant by “capacity” anyway!
Daily capacity means per 24 hrs (and implies 24 hrs of continuous
operations) and is more or less unambiguous…except that you don’t
know whether the cited number is the manufacturer’s rating on the kiln
(usually a conservative figure), or actual “full blast—but normal
operating—conditions. Commonly, the rated capacity is lower than the
actual operational capacity. Annual capacity is not well defined. The
USGS calculates an “apparent annual capacity” as follows:

Apparent annual capacity = (Reported daily capacity) * (X-Y) where:
capacity is in tons; X is 365 days (or 366 in leap years) and Y is the
number of days of downtime for routine, scheduled, maintenance. Thus,
(X-Y) is the scheduled operational year. The PCA, I believe, uses the
same formula, but (in my opinion) does not always check to distinguish
routine maintenance downtime from total downtime. Clearly, there can be
a large variation in apparent annual capacity if there is any
mischaracterization of the routine maintenance downtime or the daily
capacity. The USGS calculates—for the plant overall—the capacity
utilization wrt. the actual clinker output: we query (question) any
utilization percentages in excess of 100%.

Because there is commonly downtime in excess of routine maintenance
(extended upgrades, shutdowns for unplanned repairs…), the USGS feels
that a utilization rate of 85% or more represents full practicable
capacity operations.

[The industry reports to the USGS, for each kiln, the daily capacity,
the total downtime, the subset downtime for routine maintenance, and the
remainder for “other downtime”. We also ask for the dimensions of
the kiln, whether it is wet or dry, whether it has a preheater, whether
it has a precalciner, and whether dust control is via a baghouse or an
ESP.]

p. 1-6: Figure 1-3: the color scheme would be improved by using blue for
wet and red(dish) for dry. Cite the data source.

Para 1 below Figure 1-3: Here’s where you are tripping over your
misuse of the term “Dry” kiln. First, our table 5 did show 16.3%
from wet plants, but this is NOT quite the same thing as saying that wet
kilns did 16.3% of the clinker. This is because there is a separate line
for the relatively small output from “combination” plants (those
that operate both wet and dry kilns). Likewise, the 11.8 Mt of clinker
form Dry plants understates the output of the dry kilns because a couple
of these are in the line for combination plants. And you misstate the
11.8 Mt “Dry” as being from dry kilns lacking a preheater or
precalciner. Not so—the datum is for dry plants of all 3 types (long
dry, preheater kilns, preheater/precalciner kilns).

The USGS, not the PCA, should be cited as the source of the value data
and the consumption numbers. For 2005, portland cement sales to final
customers (these are the data that the PCA calls “consumption” were
worth $11 billion; masonry cement sales $0.68 billion; total about $12
billion.

Cite the USGS also for the statement that the top 10 cement companies
had 80% of the production; ditto for all production by State data.

Figure 1-4: this plant location map is incomplete because it fails to
adequately discern localities with more than one plant in a cluster. For
example: WA has 3 plants (2 integrated plants @ Seattle and a grinding
plant at Bellingham); CA has 11 plants; FL has 6 plants, etc…

p. 1-7: your discussion of imports is misleading: imports can be by
domestic producers seeking to feed markets where domestic (particularly
their own domestic) production is inadequate (or where they wish to
capture market share) AND by independent importers who compete against
domestic production in some markets, and help feed markets not easily
serviced by domestic production.

Oddly enough, there is a lot of flexibility in how the industry views
the cost of imports vs. cost of local production. A company might
continue to import—even at a loss or minimal profit—for at least a
while if this is the only way they can keep their customers happy (i.e.,
not going to another supplier).

The existence of imports can affect where domestic production is
targeted for marketing. A good example is the Los Angeles basin: a big
chunk of the demand there is met by imports, which then free up a lot of
the actual S-CA production to be sent into the Las Vegas market and, to
a smaller extent, into Arizona.

Table 1-3: The data in this table represent U.S. Census trade data (as
supplied to and then compiled by the USGS). The original data were in
metric tons. DO NOT CONVERT THESE TO SHORT TONS!!! Also, as best I can
tell, the data are for imports of “hydraulic cement and
clinker”—meaning, all types of hydraulic cement, and all types of
hydraulic clinker--so they are not just portland cement imports. This
(cement and clinker) issue applies, I think, to all import data that you
use in this report and represents a fairly serious problem for your
model.

The percentage column needs clarification: percentage of what (USA total
imports).

Figure 1-5 needs a citation for the data source (USGS I presume). The
curve for import share is misleading and represents a problem with the
USGS data: we compare the level of cement imports to that of total
consumption (apparent consumption calculated as cement production +
cement imports – cement exports – change in cement stocks). We
exclude the imports of clinker because the production of cement includes
cement made in the USA from imported clinker. But we do not include
changes in clinker stockpiles (but should) mainly because we did not
collect such data prior to a few years ago. Done “properly” the
import reliance would be 1-2% higher than what we show currently in the
data series.

Another related issue is the cement stockpiles: the comparison is
between successive yearend (12/31) stocks. But you could have a
significant drawdown or buildup due entirely to the early arrival or
late arrival of a few ships—this would cause a stockpile shift
entirely UNRELATED to any economic forces. So I do not find apparent
consumption to be a very useful statistic.

Sec. 1.2: Emissions

Para 1: a bit clumsy explaining where S comes from—it is from
sulfur-bearing compounds in both the fuels and the raw materials. Many
organic compounds (in fuels) contain sulfur, as do kerogens in
sedimentary rocks like limestone. Both fuels (especially coal) and raw
materials can contain sulfide minerals (chiefly pyrite or marcasite);
and sedimentary rocks may contain sulfates like gypsum.

If you are going to quote the total weight of PM10 emissions, put it in
the context of something—such as the weight of the total raw materials
and fuels.

Figure 1-6 is small to the point of uselessness. Why not make it into 3
Figures (1-6a, 1-6b, 1-6c)?

The text in this area ought to mention CO2. In fact, the whole section
is awfully brief—but perhaps you could refer the reader to the
discussion in the Appendix.

Sec. 1.3: ISIS

Figure 1-7: should you add (flowsheet box for emissions) HAP and PM? 

References: I am always a bit nervous when I see the USGS Mineral
Commodities Summaries cited—these are “quick & dirty” summaries
that are meant to be just that. We prefer, where possible, that data be
obtained from the Minerals Yearbook, as this is a MUCH more
authoritative publication.

Ch. 2: Data used in ISIS

2.2.1.1: The USGS has a kiln count of 183, not 181 (=PCA number?)—but
the difference is likely kilns active during year (183) vs still active
at yearend (181).

Figure 2.1: I suppose that your market grouping has its virtues, but
there are some very strange groupings re. the cement plants. For
example:

Why on Earth is Ash Grove, Leamington, UT grouped into Phoenix, but
Holcim Devil’s Slide is Salt Lake City? Let me assure you that the
Leamington plant feeds markets primarily in UT (incl. SLC) and NV (Las
Vegas) (plus v. minor tonnage, only, into CO and NM; nothing into AZ).
In Las Vegas it competes with the S-CA producers. Why is Ash Grove,
Durkee, OR grouped with Salt Lake City—its market is primarily OR, WA,
ID.

Some plants straddle more than one markets: Lafarge’s Sugar Creek, MO
plant can indeed feed westward into the Kansas City area, but it also
ships a lot of cement down the Missouri River (i.e., eastward).
Buzzi’s Cape Girardeau plant probably has the widest geographic
dispersal of its cement than any other plant.

Overall, there are all sorts of flows up and down, in and around, the
entire Mississippi-Missouri-Ohio River system. Not sure some of these
market breakouts in the midcontinent make any sense.

2.2.1.3: The 2005 demand for portland in 2005 was 122.4 Mt (million
metric tons), and for portland + masonry, it was 127.9 Mt. Thus I
question the 127.6 portland number in your report.

I would urge extreme caution in using the PCA demand forecasts. First,
they change them at least quarterly. Second, it is likely that you have
used reports predating the recent collapse of the economy. Cement sales
fell nearly 10% in 2007, another 10-11% in 2008, and will likely drop
around 15% in 2009. The PCA has revised its forecasts downwards.

The fall in sales in 2007 was accommodated by a huge drop in imports and
just a small drop in production. There was less flexibility in
2008—imports continued to plummet, but production has fallen
substantially, as well. Both will likely fall further in 2009.

Secondly (and related), we have had plant closures in 2008 (I think 3)
and announcements of about 7 more in 2009, and all/most but the most
advanced expansion/greenfields projects are now on hold or have been
cancelled. American Cement’s (FL) brand new plant is just sitting
there; the kiln as yet unfired. Sumter Cement (FL) is on indefinite
hold. Votorantim near Macon, GA will likely never materialize. Titan in
NC will likely be delayed. Cemex at Seligman, AZ was probably never more
than a bluff, anyway. Ash Grove at Moapa (Las Vegas) was cancelled a
year ago.

Of the closures, it is as yet unclear how many are permanent—but the
closures include Buzzi’s Independence, KS plant and Fredonia (I
think), St Marys Dixon, IL, plant, Holcim’s Dundee, MI and
Clarksville, MO plants (both are wet plants), Cemex @ Brooksville, FL;
TXI at Crestmore, CA and, temporarily, at Oro Grande; Cemex at
Davenport, CA (likely temporary); some of the kilns at TXI (or was it
Ash Grove) @ Midlothian, TX; Essroc at Frederick, MD (no surprise--is
due to upgrade project at their nearby Martinsburg, WV
plant)…etc…And lots of plants are taking extended maintenance
shutdowns, which augurs for lower capacity utilization percentages in
2008 and 2009. So any forecasts of capacity growth would be of dubious
merit, at the moment; you need to use caution in relying on the PCA
forecasts.

Table 2-3: put the data back into mt. Data include clinker. These are
U.S. Census trade data (the USGS just compiles them into convenient
tables).

p. 2-5: Cement plants can certainly be older than 50 years! There is a
definitional issue. You could have had a cement plant on site for nearly
a century, but it may have been upgraded along the way. Generally, you
wouldn’t build a plant on a site having < 100 years of limestone
reserves. So the age can be a “site age” or an “equipment age”.

Avoid like the plague the use of the word “endogenous”. The phrase
“…ISIS endogenous capacity growth can also lead to endogenous
retirement of kilns.” conjures up all sorts of imagery unrelated to
cement…Couldn’t you just say something like: “capacity expansion
projects can include the retirement of existing kilns…”?

p. 2-6: “…such as coke…” Be sure specify petroleum coke
(petcoke). Our old tables showed for many years just “coke”, then we
attempted to differentiate petcoke from metallurgical coke (i.e., from
coal) without much success (not sure the respondents understood the
difference). Likely that all coke use (outside of major
blast-furnace/steel complex areas) has been of petcoke, and I am not
sure that metallurgical coke has ever been in common use even in the
steel areas—it’s too expensive.

Figure 2-2: awfully crude breakout of fuels. Where’s the petcoke (in
with the coal???). What is “Petroleum”---do you mean petroleum
products? Crude petroleum? If the former, do you just mean distillate
and/or residual fuel oil, or do you include all sorts of waste oils and
other petrochemical waste fuels? Where are the alternative fuels?

Also, do the percentage data refer to tonnages or to the heat
contribution (Btu)?

If I use actual industry-reported heats, my own data for 2005 show:

Natural gas (3.72% total Btu); fuel oil (0.76% [seems low]); coal
(64.1%); petcoke (19.3%); metallurgical coke (nil); tires (3.2%), other
solid wastes (0.5%); liquid wastes (8.4%).

Your report cites total heat (EIA) of 451.2 trillion Btu. My (USGS) data
for 2005 has 366.7 trillion Btu excluding electricity, or 375.4 trillion
Btu if I impose some “standard” heat conversions for the fuels.
Using actual data, if I include the electricity, then we jump to 413.4
trillion Btu. Thus, I suspect that the EIA data are either too high or
that they, at the very least, include electricity. Based on the rest of
your report, I do not think that you should be converting electricity
(kWh) into Btus; use the Btus just for the fuels burned at the cement
plant. That is, skip the electricity (in your heat analysis).

Table 2-4: shouldn’t these data match those of table 1-1 (or vice
versa)?

Important: as you may realize, the PCA data on unit consumption of heat
(Btu/mt—and be sure you know which ton they are using!) generally are
based on an “equivalent ton” which is a bit of an apples & oranges
mix. At one point, years back, I figured out why they use this, but I
have forgotten and they seem to have forgotten as well. You can
recalculate to a per-ton-of-clinker basis, but I do not know if you have
done this in table 2-4. 

Also, I don’t recall the PCA’s Labor & Energy surveys mentioning gas
flows—so it is unclear if the NM3/kg data are from the PCA or some
other source. PCA 2004 (Bhatty et al.) is a huge tome (very good book!),
so you should cite the actual page(s) from which the data are taken.

Table 2-5: define “mg” (milligram or microgram?)

See my earlier comments on SOx. Regarding NOx, I thnk you should note
that it is extremely difficult to get precise data on NOx outputs, given
that output varies (wide swings) minute-by-minute. But your averages for
NOx and SOx are reasonable.

While I agree that raw materials play a significant role (and hence
there can be regional trends) in the SOx emissions, it really applies to
the potential emissions (actual emissions also vary with scrubber
technology, if any, kiln technology, and, of course, the fuels). Coal is
hugely important—is the plant burning a low- or high-S coal?

p. 2-8. Your calcination emissions factor of 0.53 t CO2 per ton of
clinker is slightly too high. Instead of adopting this Andover
Technology default (see my comments on Appendix A) you should use the
IPCC default (which is what the EPA usues for the USA GHG Inventory. The
Andover Technology average CaO content for clinker is a bit too high.

So use 0.51 or 0.52 (the latter includes a 2% addition for “lost”
[not returned to kiln] CKD).

The average CO2 emissions per ton of fuel noted in this section should
be rounded! Remember, fuel consumption data are not all that precise,
and one is never told if the reported weight is on a dry basis (as in
lab-dried), standard dry basis (i.e., actual dry feed to
kiln—typically has several % moisture), or is some wet slop at an
initial handling point. Further, one can ask for—but the reporting may
not always be—as gross or high heat. You may be getting net or low
heat.

These are among the (many) woes of data collecting!

Table 2-7: I’m not sure you’ve made clear why you are showing H2O
production. 

Table 2-8: low-NOx burners are great, in theory, but lots of plants
report problems in their actual use (have lower clinker output when
using these burners).

Ch. 3: ISIS Mathematical Framework

p. 3.2: the factor: produ_r(t,r)/0.92 has a clinker ratio
(clinker/portland) of 0.92. This is too low for a good average. Even
including the blended cements and masonry cements, the USGS data show
the ratio to be right on at 0.946-0.950 (you can’t ignore these
cements because the USGS raw materials data is for portland & blended
cement). Anyway, 0.92 implies 8% gypsum and/or other additions, and this
is too high right now. If the up-to-5% limestone addition becomes
widespread (current version of ASTM C-150), then 8% may become
reasonable.

We can approach a 92% clinker factor hypothetically if we assumed that
we had a 95% market share for true portland cement @ 95% clinker + 5%
gypsum, and 5% market share of masonry cement averaging 50% clinker
(this is probably a bit low): (0.95 x 0.95) + (0.05 x 0.5) = 0.9275.

Again, the import data that you are using appear to be for both cement
and clinker.

The factor imports (t,r) is likely prone to large error because we
simply do not know how much cement coming into a certain Customs
District is consumed in that district. For example, there is
considerable barging of cement along the midAtlantic and Gulf Coasts;
ditto along the NW (Pacific) Coast. One of the few places where the
assumption (of local consumption) is valid is Florida. The worst case is
with New Orleans: imports into New Orleans can reach Chicago!

And then we have the whole issue--sadly, very common--of cement swaps
among companies.

Production and Capacity Changes: the lead-in sentence is a bit unclear:
“Regional production can be from existing kilns, kilns added at a
plant (i.e., expansion kilns), newer kilns replacing kilns at a plant
(i.e., replacement kilns), and new kilns.” 

I think you mean to say something like: “…Future
production…existing kilns, upgrades of existing kilns
(debottlenecking, major technological upgrades), new kilns at existing
plants (perhaps replacing existing kilns), and new plants altogether.”

Here’s how I look at things:

Present kilns		Production?		Future production

Idle kilns:		No			Maybe will be restarted

Active			yes			yes (same capcity); and/or

						Upgrade kiln line; modify feeds (eg CemStar); and/or

						Debottlenecking; and/or

						New kiln at existing plant;

						Shut down old kiln; and/or

						New (greenfields) plant.

It is unclear if your discussion, and projected capacity changes, fully
account for plant shutdowns (now seen to be a much faster schedule than
was anticipated even 1-2 years ago). Have capacities been
properly/uniformly defined for all new kilns (are they summed on the
same basis)?

p. 3.2-3.3: Equation 3.1.3. for regional demand: I suggest using the
word “flows” instead of “trade” when referring to inter-State or
inter-market movements of domestic cement.

Equation 3.1.4 only holds if the import: total demand ratio is accurate,
and I’m not sure this is the case. Again, I would note the fact that
imports into the New Orleans Customs District feed the entire
Mississippi-Missouri-Ohio river system. How are you determining where
cement imported at location X is being consumed?

p. 3-3: the price data for imports (Equation 3.2.5.) are very weak. You
only have import values reported on a CIF basis (USGS monthly and annual
reports) and Customs Value (USGS annual reports). The CIF value does not
represent the actual price because it has not yet been offloaded from
the ship: that is, the true price is CIF + various port charges +
markup. 

And the CIF (and Customs) value data themselves are full of errors: sad
but true. Not much that can be done about this. 

Much of Ch. 3 is not exactly reviewer-friendly (!) and I really was
unable to do much with it absent detailed experience with running the
actual model.

Ch. 4: Calibration of Cement ISIS Model

4.1.3.1. I am not sure that the assumption that Priceimports = producer
price is at all valid. Dumping issues come to mind. I have found that
imports can be cheaper, the same, or higher than those from “local”
producers. One issue is transportation costs. Another may be differences
in the type of portland cement: local production might be Type I and the
imports Type III. One thing I will say is that “cheap” imports do
not necessarily equate to cheap sales prices—importers are very fond
of large margins.

You cite a VERY limited number of examples of import-dominated markets,
and without really defining what this means. What is the percentage
criteria for this designation? Seattle may not be an especially good
example.

USGS data for 2005 (MYB Table 9) show consumption in WA as 2.348 Mt and
OR as 1.237 Mt; against imports into Seattle Customs District (CD) of
1.489 Mt and 0.867 Mt into the Columbia-Snake CD. This looks like a high
import ratio, but do you really know where these imports are being
consumed? 

Again, things fall apart for New Orleans and points upstream.

Table 4-1: why 2004 data and not 2005? Again, switch to metric tons!

What is the basis of the production data in table 4-1? The USGS data do
not allow for it, as our District groups do not match your market
groupings. The PCA normally does not publish production data other than
in broad groupings (kilns > 500,000 tpy, etc…).

High import ratios exist for Florida, Los Angeles, San Francisco, San
Antonio—yet the table designates (**) only Florida and Seattle as
having high ratios.

4.1.3.2. The PCA may have predicted higher import ratios in the future,
but hey certainly have fallen in 2007-09 and likely will stay down in
2010 as well. Thereafter, I venture not.

Table 4-2: where are you getting your price data; and why are you
showing 2003-2004 and not 2005? Need to cite a data source.

Table 4-3: The emissions shown are too precise.

Ch. 5: Illustrative Analysis

Figure 5-1 and the discussion of it lead to the conclusion that the NOx
cap may be too restrictive. Whether this is valid or not, however, is
hard to determine. Apart from the difficulty for a plant in precisely
controlling NOx emissions, in typical operation, the emissions from a
given plant typically swing rather widely on a minute-by-minute basis.
Because of this, it has always been an issue as to how one is to
determine the actual emissions volumes over a specified period, as it
could affect whether a plant exceeds its NOx cap or not. I think this
report needs to discuss this.

In contrast, Figure 5-2 would seem to illustrate the relative ease in
controlling SOx.

Figure 5-3: Figures should stand alone—thus it is confusing why the
emissions caps and banking are not shown (I know the answer—just, for
now, leave these off the caption).

I do not know if the projections in Figure 5-3 have made allowance for
clinker imports and changes therein. 

Given a likelier harsher reality re. plant closures, these projections
may need to be redone.

5.1.2.: In the text, it is important to minimize the use of acronyms,
especially for readers not conversant with these technologies. The
savings of space are not that much!

One important issue with CemStar is that steel slag may not be available
locally.

Figure 5-4 is impossible to read for those of us equipped with standard
or substandard human eyesight. After staring at this long enough, I am
left wondering as to why there is such a huge drop in LNB + TDF in 2011
wrt. 2010. Are the subsequent (2011--) years in addition to the 2010
levels? (i.e., are they just shown as margins?). Was their a scale
change post-2010?

5.1.3: The total demand in the near term has now likely been revised
downwards by the PCA.

Figure 5-5: it is not immediately clear why the red dot has been
displaced rightwards (I can guess, but it looks strange). And what
happened to the red dot for 2025?

Looking at this figure, I do not see much difference between the base
case and the policy levels.

Figure 5-6: Unless I am missing something, the blue dots are not needed
at all—the sum is always the top of a stacked bar graph. The Legend is
confusing, as I only see one market that has either a trade-in or
trade-out (and, again, I would urge calling this an inflow or outflow)
and that is Birmingham. Surely, there are more intermarket flows than
that! And the data for imports are either essentially all missing or you
are projecting zero imports. I do not understand how you have determined
the (foreign) imports into the various markets—such as Dallas. The
trade data for that area are under the Houston-Galveston Customs
District, which is in your San Antonio Market area (yes?). Why are no
foreign imports shown for Birmingham? Florida???, Phoenix?? St. Louis?? 

So this figure needs some work….

5.1.4. the cost analysis seems to show that, of the total increase in
costs ($1.58 billion), all but $0.01 billion is due to control
technology. What about increased fuel costs?

Ch. 6: Summary

Basically OK. Perhaps you need to again mention here here why CO2
emissions could drop under SO2-reduction strategies (somewhere earlier
it was stated that there could be decreased CO2 using limestone wet
scrubbers (LWS)—it is unclear why this would be so. I would think that
reduced CO2 would instead be—as stated elsewhere—from changes in the
fuel types that might occur in a SOx-reduction program.

Appendix A: Staudt, Jim, Andover Technology Partners, “NOX, SO2, and
CO2 emissions from Cement Kilns”:

This is a rather good report with a few places that bothered me.

Table 1: the exit flow gas rates would show a lot of variation; not sure
how “good” an average these rates are. By the way, Bhatty et al.
(PCA 2004) cited is a huge volume, so a page reference would have been
helpful.

Table 2: although the heat values shown are reasonable, it should be
pointed out that, in reality, most fuel categories have a range of heat
values, and that it is NOT guaranteed that the coal burned by plant X in
year Y will be the same coal as burned in year Z. Further, the reporting
of fuels by cement plants has problems, among which are:

who is reporting: a lab technician, a plant manager, a
controller/accountant? Same person each year?

is the weight of (e.g., coal) the actual weight going into the kiln? In
it expressed as the actual mass or converted to a constant moisture (>
0%) content basis or to a zero moisture basis? 

Is the heat value provided or not? If so, is it a gross (= HHV) basis or
net (LHB) basis?—the former is preferred for emissions calculations. 

Is the heat value provided as a Btu/st (or other unit) ratio (USGS then
multiplies it out), or did the plant report the full Btus? If the former
(Btu/t), was it “real” or simply calculated by the plant: tons x
Btu/t = Btu – and if so, was the same Btu/t used for all the coal used
during the year, or were periodic checks/adjustments made to the Btu/t? 

for whole-tire fuel, the moisture content can vary dramatically.

Have any waste fuels been identified properly: if no heat was provided,
how do you impose a heat estimate for “liquid waste?”
(undifferentiated). How about for an identification given simply as:
“used solvents?” Or “off-spec oil and lubes, inks, and
solvents”. What about things like: “(solid wastes: oily rags, street
sweepings; contaminated soils)”. Or even “high-carbon fly ash”?

These are significant problems because the carbon factors for fuels are
generally expressed as “ amount of carbon per X amount of heat”

Anyway, I would also comment that the values in Table 2 are expressed
with more precision than is warranted.

From table 2: coal at 12,456 Btu/lb becomes 27,460,772 Btu (HHV)/mt and
26,340,824 Btu (LHV)/mt. USGS data (asked for as HHV) typically show a
range of 22-28 MBtu/mt. So table 2 appears to be at the upper end of the
range.

CO2 from calcination: Although Staudt’s calculation is arithmetically
sound (within rounding parameters), I do not like the default he derives
(0.53 t CO2/ t clinker) because it is based on an average CaO content of
clinker of 67.6%. This is too high—it is near the upper end of the CaO
range in modern clinkers. As was exhaustively dealt with by the IPCC,
the average is a bit lower. No great precision is justified owing to
variation in the chemistry of clinker at various plants and among
plants, but 65% is a better, more useful average. This drops the default
(using the same assumptions as to CaO source) to 0.51. If you
incorporate the IPCC-suggested 2% additional CO2 for CKD “lost” (not
returned to the kiln) to the system, it becomes 0.52%. Neither protocol
adjusts for MgO contributed from a carbonate source, but this is OK
because the assumption that the CaO is 100% from CaCO3 already causes a
compensating overestimate; MgO is always kept low anyway (0.5-2%; seldom
higher), and MgO is commonly from silicate phases instead of/in addition
to a carbonate phase.

In any case, the EPA in the USA GHG emissions inventory uses the IPCC
default (w/ CKD) of 0.52 and I would suggest that the ISIS model do the
same.

It should be noted that the CaO content of the actual “limestone”
burned by the cement industry varies a lot, both across country and even
within some individual quarries. Staudt has been incautious in equating
“limestone” and calcium carbonate (calcite): the former is a rock
mostly made up of calcite; the latter is a mineral of formula CaCO3. His
stoichiometry applies to calcite or calcium carbonate, not to limestone.

least efficient → 4 most efficient) as follows: total heat (total fuel
consumption): 1 wet→ 2 long dry → 3 preheater → 4
preheater/precalciner (precalciner). Electricity: 1 precalciner → 2
preheater → 3 long dry → 4 wet. But we see other patterns: for two
plants of 1 Mt/yr capacity where each has similar technology kilns, a
plant having a single kiln will be more efficient than the plant with
multiple kilns. This relates not only to economies of scale but, for
electricity, on the fewer motors needed for the single kiln.

p. 8: while it is true that, to some degree, flame temperatures in
precalciners may be a bit lower than in the kiln tube itself, the
different proportion of NOx type (mostly thermal NOx in the kiln tube;
mostly fuel and/or feed NOx in the calciner) also derives from the fact
that: 1) the burner for the calciner generates a shorter flame, which
may reduce thermal NOx, and the calciner is dealing with higher surface
area particles, so does the calcining far more efficiently than the kiln
tube could do. Thus there will be less NOx overall for the calciner on a
per ton of clinker basis. And, because of this efficiency, the kiln tube
itself will be smaller on a precalciner kiln line, which makes for
better efficiencies all around (for example, reduced heat loss through
the kiln shell).

The higher NOx/MBtu for wet kilns mentioned by Staudt may also be
related to the fact that wet kilns these days almost invariable burn a
relatively high proportion of waste fuels. Also, there could be some
correlation with things like the use of primary vs secondary air. And,
many of the wet plants operate multiple kilns. And, being older, and
larger tubes, with more motors/tires (hence kinking), there may be more
frequent stoppages, hence more use of warmup fuels (natural gas, for
example, makes a lot of thermal NOx because of the exceptionally high
flame temperature). Finally, the wet kiln plants may have older NOx
monitors and maybe these are skewing data (???).

I agree with Staudt that the NOx data show a lot of scatter. I would
note that his observation that the larger kilns are more efficient
(lower NOx) is perhaps not supported by his graph—yes it shows this,
but there are only 3 data points for the kilns > 1 Mt/yr capacity. Is
this statistically valid?

Figures 5a-5c: Author is being “kind” in his characterization of the
data: the correlations would appear to be not statistically significant.

In his discussion of SOx capture, the author says that synthetic gypsum
is useful to the cement plant (saves them buying outside gypsum). Yes
and no. In realitity, some syngyp is dirty. A common complaint about its
use in the finish mill is that it is sticky. So far, not that many
plants are using it (but, in fairness, they are not required to split
syngyp from natural gypsum in their raw materials reporting to the USGS,
so we may not be seeing the full extent of syngyp use).

p. 15: perhaps this is a semantic point, but Dragon Cement’s switch to
dry technology involved shortening the existing kiln tube, not by
bringing in a new tube.

While it is true that there is regional variation in the sulfur content
of “limestone” (and the sulfur in “limestone” is from 3 main
sources: sulfide minerals such as pyrite; kerogens; and
gypsum/anhydrite.), there is as much, or greater variation from the
S-contribution from the coals: there is a major difference in the
S-content of coals (S in coal is from kerogens and sulfides like pyrite
and marcasite).

The sulfur content of the “limestone (BTW—Dragon actually burns
marble; and much of the limestone burned in N-CA is also a marble, or
nearly so) would not normally be a criteria for selection of a new
quarry.

Given the small number of kilns in most of the markets, I am not sure
that one ought to read too much in the mean vs. median emissions data in
tables 8 and 9. 

Appendix A continued: Re: Costs and Performance of Controls

This is also a good report.

Regarding the use of tires, I would comment that not all plants burn
tires because:

there can be local competition for tires—you need an “abundant”
supply, not just the occasional tire;

permitting issues

the requisite tire supply may not be available in all areas (not near
enough to a large city);

Capex for the tire-burning and feeding equipment.

I am not sure that I agree that tires yield slightly less CO2 than coal:
perhaps on a dry weight basis they yield slightly more. Depends on how
you relate the emissions: per weight of fuel, per ton of clinker…

I agree that tires are not fossil fuel, per se, but most involve a
considerable input of the products of the petrochemical industry in
their manufacture.

p. 13: Preheaters and precalciners are described as “post-kiln”
combustion technologies. I guess that they would be if you are looking
up-kiln (i.e., from the flame and gas flow directional perspective), but
most people would follow the raw materials and clinker instead and would
call preheaters and precalciners: “pre-kiln” technologies (that’s
why “pre-“ is in their name!).

I think most calciners are designed to burn about 60% of the (total kiln
line) fuel, not 50%; though perhaps the oft’ quoted 60% includes riser
duct firing.

As stated in para 1 of p. 13, there is an implication that most thermal
NOx is avoided in preheater and precalciner kilns—as stated, this is
incorrect. First, there is a lot of thermal NOx generated in the kiln
tube from the main burner. Secondly, a preheater kiln commonly uses hot
combustion air for the preheating (which is full of thermal NOx but this
may be destroyed if reducing conditions exist in the preheater). What is
true is that NOx generated in the calciner (and hence also carried into
the preheater cyclones in a preheater-precalciner system) is dominated
by fuel and/or feed NOx. So the issue is that the precalciner (location)
doesn’t generate a lot of thermal NOx—the same cannot be said for
the whole precalciner kiln line.

p. 26: LWS scrubber: basically fine. Here and elsewhere, use calcite, or
calcium carbonate, instead of limestone. Limestone is a rock, only very
rarely is it pure calcium carbonate.

∙2H2O .

p. 30: CKD is described as being made up of oxides of Ca, K, and Mg.
You’ve left out Si!

Para. 2 states that CKD disposal is not an issue with dry-technology
kilns. This is not true. Any kiln technology can burn CKD and any can
have a problem burning CKD—it depends primarily on things like the
alkali content of the CKD and the resulting clinker. If there is a
regional ASR problem with the local aggregates (in concrete) then you
need to keep the alkali content of the clinker down. This may be
well-controlled if you have an alkali bypass system, but if you do not,
then your ability to burn CKD may be severely limited.

You can certainly burn CKD in a wet kiln: the slurry viscosity can be an
issue, but you can also add a bit more water. CKD was burned at
Dragon’s wet kiln, for example, but only in relatively modest amounts
(they did not burn all of it).

CemStar: I would add a bullet for the CO2 reduction!

On the negative side, CemStar may add to a plants emissions of Cr+6. I
think this is why Cemex’s Davenport (N-CA) plant is presently idle.

Good for you in getting the $16/ton royalty figure: I am in the business
of collecting proprietary data and I never could get anyone to tell me
this number (I heard grunts and mutterings about it, it was high; I
heard $12 mentioned once but it was not definitive). The high royalty
definitely limited interest in the technology. But, there is an
important difference between what TXI got/continues to get for the
CemStar contracts that were signed with TXI, and what the slag
processing company Levy (who bought the CemStar patent in 2004) gets for
contracts it develops. Levy is primarily interested in promoting the
slag use (and selling the slag to the cement plant) and charges (they
tell me) a much lower royalty (but they don’t tell me what it is). My
point is that, in your cost analysis, it is likely that future (new)
users of CemStar will be paying <<$16/ton royalty (if any) for the
process.

At $16/st, CemStar becomes quite expensive relative to the cost of the
steel furnace slag (a few dollars per ton only, + transportation) and
its handling.

I do not understand this $72.59/st cost for clinker (seems excessively
high), nor why it is even relevant. If I understand the report
correctly, it is being said that this clinker cost minus $10/ton for the
slag, yields a savings of $62.59/ton or $45-50/ton including the CemStar
royalty. I just don’t see how this relates to evaluating the use of
CemStar. What you seem to be doing is pretending that you stick slag
into the finish mill to make cement (that it competes with clinker)
when, in fact, it is “competing” with (some of the) limestone as a
feed to the kiln.

What you need to do is to compare the cost of making clinker without
using CemStar vs the cost to make clinker using CemStar. CemStar always
involves incorporating steel slag into the raw mix as a partial (3-10%)
substitute for limestone (etc…). 

Without CemStar: cost = A (cost to quarry, comminute, & blend the raw
feed materials) plus P (the cost of pyroprocessing (fuel costs etc…)
to make clinker. This is the cost to make the clinker, period. 

With CemStar: for every ton of clinker made, you are quarrying,
comminuting, and blending less “normal” feed, so A is lower by an
amount B. But, against this, is the cost (A’) of: buying the slag, the
royalty for the process, crushing the slag to 1-2” size, and blending.
So the resulting feed cost becomes (A-B + A’).

The pyroprocessing cost is the cost of calcining the original feed mix
(in the now smaller amount), plus the cost of melting/disassociating the
slag, plus the cost of sintering. Because there is less carbonate to
calcine (per ton of clinker), the fuel (heat) required for calcination
will be less. (Calcination is where 60% or more of the heat is
consumed). Also, because the slag melts/disassociates easily, you are
getting some of your silica, iron, and aluminum (oxide) for less heat
(and maybe faster) than would be the case burning clays, shale, and
quartz sand. So the cost savings are related to the reduced heat/fuel
charges of the pre-sintering stages of pyroprocessing (P’). For the
sintering part of the clinker-forming process, there probably will not
be any savings. So the pyroprocessing cost becomes P-P’.

So the overall comparison is A + P vs. (A-B + A’ + P – P’). And
the overall savings will be nowhere near $45-50/ton of clinker! 

The reduction in CO2 emissions will be, on a per-ton-of-clinker basis,
will be something like: for calcination emissions: (1-C’)*C, where C
is the calcination CO2 for the carbonates burned in feed that does not
incorporate CemStar, and C’ is the percentage of carbonate
substitution by slag. That is, 5% (C’=0.05) substitution yields:
(1-.05)*C = 0.95C. i.e., this reduction is stoichiometrically
straightforward. The fuel emissions reduction is a bit less so. If F is
the non-CemStar fuel emission for the pre-sintering stages, the new
emission (w/ CemStar) becomes something like: ((1- C’)*F) + G + S
where the first term relates to the reduced heat of calcination and
low-temperature pyroprocessing of the normal feed mix; G is the low
temperature fuel requirements to deal with the pre-sintering
disassociation of the slag; and S is the fuel requirement for sintering.
I am assuming that S does not change (i.e., fuel emissions without
CemStar = F + S). The CemStar savings on fuels stems from the fact that
G << C’*F.

What this all means is that the statement on p. 36 that the CO2
reduction (overall, on a per ton clinker or cement basis) will be
directly proportional to the percent slag substitution is not quite
safe—it will hold for the calcination part of the emissions, but may
be different for the fuel side of the emissions.

p. 35: Again, if I am reading this correctly, it appears that a claim is
made that CemStar increases the electricity consumption by the finish
mill. Well, if you assume that you get 5% more clinker for a 5% CemStar
(slag) introduction, then, yes, it will take 5% more electricity to
grind that extra clinker. But so what? This is like saying that if you
get a salary increase, you will pay more taxes even though the tax rate
doesn’t change. This is not interesting. What would be interesting is
if the use of slag (CemStar) made a clinker (perhaps a harder clinker)
that was costlier to grind (more electricity per ton). But I do not
think that this is the case: the cost (either per ton of clinker or per
ton of cement) to grind the clinker into cement does not change as a
result of using CemStar. There may be some additional electricity in the
raw mill end of the plant to handle the extra ingredient (slag) but this
is likely to be more than offset by the reduced electricity costs of
comminution of the feed: unlike the limestone and other feeds, the slag
does NOT need to be ground, just coarsely crushed. [the CemStar patent
“discovery” was not the chemical contribution to the clinker-making
process. It is the fact that you don’t need to grind it—only crush
it to 1-2” diameter. The utility of slag had been known for many
decades, but it had not been popular earlier because it was assumed that
you had to grind it like everything else, and slag is indeed hard
(costly) to grind].

p. 37 Conversion from wet to dry

The wording is a bit imprecise. You can convert a plant from wet to dry
by replacing wet kilns with dry kilns and/or by converting wet kilns to
dry technology. Dragon Cement in Maine was a true kiln conversion: the
old wet kiln tube was retained (but about 1/3 of its length was cut
off), and a preheater/calciner tower was added. Holcim @ Holly Hill, SC
was a replacement of the 2 wet kilns (which were shut down) with a
completely new, unrelated, dry kiln (precalciner kiln).

Regarding the 300,000 ton example—I would only comment that one would
never build a kiln of such small capacity today, nor would it be
worthwhile anymore to convert a 300,000 t wet to a 300,000 dry kiln.

p. 38: states: “Water consumption would also drop. Moisture added to
the materials of a wet kiln

amount to about 0.75 tons per short ton of clinker.” This is too high
(too much water). Wet and dry kilns both need about 1.6-1.7 tons of raw
materials per ton of clinker. Typical water content of wet kiln slurries
is 35-40% by weight. So a 35-40% water slurry containing 1.6-1.7 t of
raw materials (destined to make 1 ton of clinker) would contain 0.56 –
0.68 t water.

But, of course, water is also used in cement plants as a motor coolant
and in the finish mill (as a spray) to make sure the calcium sulfate in
the cement remains (mostly) gypsum and does not dehydrate to plaster or
to anhydrite.

p. 42: The question is asked why haven’t all the existing preheater
kilns been converted to preheater-precalciner technology? There are at
least 3 answers to this:

if the preheater kiln is small (capacity), the conversion boost in
capacity might be insufficient to justify the cost;

significant projects like this typically require a litmus test of the
plant still having at least 50 years of limestone reserves. 

significant projects like this likely will put the plant into a stricter
regulatory framework re. emissions (the existing plant may have
grandfathered in).

General comment: One significant omission from this study—unless I
missed it somewhere (it is a long report!) are certain upgrades to
capacity that simply involve debottlenecking. This could involve things
like:

upgrade/replace clinker cooler

upgrade raw mill

upgrade finish mill (e.g., install roller mill instead of ball mill—it
will greatly reduce electricity consumption)

Upgrade storage facilities (raw feeds, clinker, cement) and overall
materials handling.

Upgrade dust collectors

p. 44: don’t need to capitalize greenfield(s) and brownfield(s).

p. 45: Not sure that “attainment” status is clear to all
eaders—may be environmental jargon.

Item #2 (reserves). 100 years for a greenfields project is about right.
To this I would again note that major capex projects need c. 50 years of
remaining reserves or more (some companies may be satisfied with 35
years).

Item #3 labor: I do not think labor is important…Look, they have
cement plants all over the world. Labor can always be found. The more
skilled positions will draw engineers (et al.) from far away. When I
have visited cement plants, I am commonly struck by the fact that most
of the workers have been there for ages—they have had or are planning
on having their whole career there. They evidently like the work. The
pay is adequate and steady; it’s relatively clean, and relatively
safe. It’s a more recession-proof industry than most. So there’s not
necessarily a lot of overturn (oddly enough, except for the plant
managers). People learn the work; they get trained.

Modern plants do not need a huge workforce anyway. And, normally, you
would not cite a cement plant in the middle of the Arctic wilderness.
Unlike certain metals mines (where you simply must put the mine where
the deposit is located or else the mine will fail badly), you generally
have a choice as to where you site the cement plant (limestone is
abundant). Generally, there will be some sort of community nearby the
proposed cement plant that can supply the “blue-collar” labor.

You offer the notion that the USGS has/will map out suitable limestone
deposits. Unlikely. Cement companies have their own staff of geologists,
or will hire a consulting geologist, for this purpose (they may use USGS
and/or State geological maps to broadly identify where the limestone may
be found).

p. 46: Decision tree: add: consider adding a terminal (instead of a
plant); consider swap arrangements (to supply cement to customers
outside your own plant’s “range”.)

Round the midpoint costs to 2 significant-figures: e.g., $280 instead of
$277.

Appendix B: (memo from RTI)

p. 3-4: I would note that the USGS data pertaining to prices for crushed
stone ($/t) are for sales by crushed stone quarries. Thus, they would be
broadly applicable to prices for some of the raw materials that are
purchased by the cement company, but likely exceed what a cement company
“pays” for its own limestone. I would think that the basic limestone
cost (quarry, primary crushing, convey to raw mill) would be typically
≤ $5/st.

The average hourly wage table look a bit low for a cement plant, but I
am not sure. Need to get some actual cement plant data. My main issue is
that the typical cement plant worker has 10+ years of experience, I
would think, and so would likely draw a higher wage.

Table 6: The electricity data seem high. First, whereas the logic of the
PCA’s “equivalent tons” may be sound for heat, it makes less sence
as the denominator for cement, given that roughly 1/3 of the consumption
is in the finish mill. The USGS data relate the unit consumption to
total cement (not just portland) production. We show for 2005: wet
plants = 135.4 kWh/mt; dry plants (all types) = 139.0 kWh/mt;
combination plants (using wet & dry kilns) = 153 kWh/mt; and all
integrated plants combined = 138.8 kWh/mt. Grinding plants = 84 kWh/mt.

Table 7: the unit electricity prices may not be valid for large
industrial customers (they may get a lower rate).

Table 10: a few comments (reflecting newer info):

The Dixon, IL and Charlevoix, MI plants are now owned by St. Marys
Cement.

Hanson plant, CA is now owned by Lehigh SW

RMC plant in CA is now owned by CEMEX

Lehigh Allentown, PA plant is actually their Evansville plant

Holcim (US) now controls all St. Lawrence facilities in the USA.

Essroc’s Nazareth III plant in PA shut its kilns in 2005 (only kilns
are at Naz. I)

Lafarge’s Atlanta, GA plant shut its kiln in 2004; now only grinds
some clinker to make masonry cement, regrinds Type I into Type III
portland cement, and grinds granulated slag.

Dragon Cement was a dry plant in 2005 (the wet to dry conversion was
completed 4/04).

This completes my review comments.

Attachment H

Review Comments from

Hector Ybañez, Holcim

Transcription of comments submitted verbally to EPA during the meeting
on January 29, 2009

Page 1-3 and 1-4: Figures 1-1 and 1-2 Update cement diagram – add
fabric filters (for PM controls) instead of ESPs and simplify the
diagram with just the cement making process

Page 1-5: Statement about PC fuel efficiency in the first paragraph…
replacement of both wet and certain dry process kiln capacity with
modern kiln processes yield substantial reductions in fuel use.  This is
true in term of fuel efficiency but not in the absolute amount of fuel
use.  Newer kilns tend to be bigger kilns and are going to use more
fuel.

Page 1-7: Figure 1-4 Add import’s custom districts to the map

Page 1-8: Do these imports include imports from Mexico (imports by rail
and trucks)?

Page 1-9: Are the PM emissions (36,000 tons) emissions from the kiln or
for the whole plant (including grinding)/

Page 1-11: Figure 1-7 – Emissions as inputs to the model are baseline
uncontrolled emissions at baseline levels of production.

Holcim will provide EPA with capital recovery cost for 15-20 years old
kilns and some information on capital cost and interest rates

Page 2-3 and 2-4: Tables 2-2 and 2-3 specify the units 

k

·

û

8

®



J

K

e

f

g

i

j

k

–

—

±

²

³

µ

¶

·

Ú

ࠆᘁ䁨夔唀Ĉ䡭Ѐ䡮Ѐࡵ⸁Ú

Û

õ

ö

÷

ù

ú

û

2

3

4

6

7

8

Ž

¨

©

ª

¬



.

®

h

h

h

h

摧寓°

_H

 h

 h

 h

 h

–

 h

 h

 h

 h£K

jÜ

਀&䘋

0

6

U

 h

 h£K

 h

 h£K

 h£K

 h£K

 h£K

 h£K

/Kg for dry seems low.  Hector will share Holcim’s calculation method
at 10% oxygen to recalculate the EGFW.

Page 2-7: Table 2-5 Update with the most recent ACT data (2008 ACT) and
show data in lb/ton clinker

Page 2-8: Table 2-6 Change red color to be light green.

Page 2-9: Table 2-7 Label symmetrically, lbmolesH2O/MMBtu, H2O is
missing in some places

Page 2-10: Table 2-8 Are CO2 reductions total CO2 (process and fuel)? 
If so, express it to be total CO2.  Are numbers absolute #s or
intensities? Can sewage impacts be included? PCA has given EPA a
workbook with the fuel and power use for the cement industry.  This
workbook can be used to compare #s.  The workbook is part of the WBCSD
initiative.  Hector will find out who at EPA PCA sent the workbook data.
 

Page 2-11: Table 2-9 Are these #s fuel intensity or absolute?  We might
think that the #s are intensity related not absolute.  When going from a
wet to a PC you have more capacity, you produce more and therefore you
will have more emissions in the absolute no matter that the process is
more efficient.

Page 2-12: Table 2-10 Specify in the model that these parameters are to
be chosen by the user.

Page 5-6: Table 5-1 control cost #’s need to be updated to
differentiate between new capacity in the business-as-usual vs. policy
scenario.

Comments on Spreadsheet

Add plot on retirement, new capacity, etc.

  PAGE   \* MERGEFORMAT  4 

  PAGE   \* MERGEFORMAT  54 

  PAGE   \* MERGEFORMAT  60 

  PAGE   \* MERGEFORMAT  62 

  PAGE   \* MERGEFORMAT  64 

  PAGE   \* MERGEFORMAT  69 

  PAGE   \* MERGEFORMAT  70 

