DEMONSTRATION PM-|0 STATE
IMPLEMENTATION PLAN FOR
HAYDEN, ARIZONA
PREPARED FOR
U.S. ENVIRONMENTAL PROTECTION AGENCY
REGION IX, SAN FRANCISCO

Duane Ono, Project Officer
JULY 1987
[ULfl
 0
Q
IUU1
                                      [LIU
PREPARED BY
ENGINEERING-SCIENCE

-------
      DEMONSTRATION PMio STATE
       IMPLEMENTATION  PLAN  FOR
          HAYDEN,  ARIZONA
            Prepared for

U.S. Environmental Protection Agency
             Region IX
          San Francisco, CA

     Duane Ono, Project Officer
             Prepared  by

         Engineering-Science
      75  North Fiar Oaks Avenue
            Pasadena,  CA
            July  24,  1987

-------
                             TABLE OF CONTENTS


                                                       Page Number

Executive Summary                                           iii

1.0  INTRODUCTION                                           1.1
     1.1 Purpose                                            1.1
     1.2 Background                                         1.1
     1.3 Study Area Definition                              1.2
     1.4 General SIP Approach                               1.2
     1.5 Report Contents                                    1.2

2.0  DATA BASE DEVELOPMENT                                  2.1
     2.1 Air Quality                                        2.1
     2.2 Receptor Modeling Data Base                        2.3
     2.3 Dispersion Modeling Data Base                      2.8

3.0  CMB RESULTS AND MODEL VALIDATION                       3.1
     3.1 Analytical and CMB Results                         3.1
     3.2 CMB Model Validation                               3.9

4.0  RECONCILIATION OF RECEPTOR/DISPERSION MODEL RESULTS    4.1
     4.1 Preliminary Dispersion Modeling                    4.1
     4.2 Modeling Reconciliation                            4.1
     4.3 Summary of Model Reconciliation Analysis           4.8

5.0  BASE YEAR DISPERSION MODELING                          5.1
     5.1 24-hour Design Concentration                       5.3
     5.2 Annual Arithmetic Mean Design Concentration        5.3

6.0  CONTROL STRATEGY ALTERNATIVES - PM10                   6.1
     6.1 Emission Reductions Needed to Attain NAAQS         6.1
     6.2 Growth Projections and Banked Emissions            6.3
     6.3 Emission Reductions-Needed for 1990 NAAQS          6.7
         Compliance
     6.4 Source Specific Control Options                    6.7

7.0  SIP CONTROL STRATEGY IMPLEMENTATION                    7.1
     7.1 Control Strategy Selection Methodology
     7.2 Implementation Methodology                         7.1
     7.3 Hayden Ctonrol Strategy and Implementation         7.2
     7.4 Preconstruction Review                             7.6
     7.5 Demonstration of NAAQS Maintenance through 1997    7.7
     7.6 Annual Progress Report                             7.7

8.0  AIR TOXICS SOURCES AND EMISSIONS                       8.1
     8.1 Health Risk Assessment                             8.1
     8.2 Toxic Emission Source Identification               8.3
     8.3 Control Options for Toxic Air Contaminants         8.3
     8.4 Further Studies                                    8.3

-------
                                                       Page Number

9.0  REFERENCES                                             9.1

APPENDIX
     Appendix A - NEA Source Apportionment
     Appendix B - Emissions Inventory

-------
                             EXECUTIVE SUMMARY

     This report describes the use of EPA's PMm SIP Development Guideline
Document in the preparation of an example PM^g SIP for the Hayden, Arizona
non-attainment area.   The Guideline Document emphasizes the use of receptor
models in conjunction with dispersion models to identify source-receptor
relationships and subsequently to develop and test control strategies for
NAAQS attainment.

     The Hayden, Arizona area contains two copper smelters, one of which is
currently inactive, and a large tailings pile associated with the ore
concentrator.  Also associated with the smelter are copper ore unloading,
crushing and- conveying facilities.  Streets and roads in the Hayden area are
generally paved, with the exception of two moderately traveled roads within
the town of Hayden.  The 40CFR81.303 TSP nonattainment area designation of
T5S, R15E remains unchanged for
     The Arizona Department of Health Services (ADHS) has operated a SSI
sampler for PM^g at the Hayden Jail site since the beginning of 1985, and
has operated a TSP monitor at that site since before that time.  To collect
data for receptor modeling, the ADHS operated dichotomous samplers at both
the Hayden Jail and Garfield sites from late November, 1986 through January,
1987.  Monitoring data indicated exceedances of both the annual arithmetic
mean (AAM) PM^g NAAQS concentration of 50 ug/m^, and the 24-hour maximum
PM^o NAAQS concentration of 150 ug/m^.  The maximum 24-hour PM^g
concentrations measured were 243 ug/m^ and 236 ug/m^ at the Hayden Jail and
Garfield sites, respectively.  The maximum annual arithmetic mean
concentration measured at the Hayden Jail monitor was 80
     For the receptor modeling data base,  X-ray Fluorescence (XRF) analyses
of filters from both monitoring locations  and of source samples collected
from suspected, major contributors provided most of the information needed.
From these analyses and source characterization data from the literature,
Chemical Mass Balance (CMB) calculations were performed for selected days of
the approximate two month sampling period.  Sample day selection criteria
included wind speed and direction and days of highest concentrations.
Statistical techniques were applied to the CMB results for model validation.
The CMB analysis provided a generally good characterization of the
contributions from the major PM^g source groups in the Hayden area to
monitored ambient levels.

     A point and area source gridded emissions inventory for the Hayden area
plus local meteorological data provided the data base for preliminary
Industrial Source Complex Short Term (ISCST) dispersion modeling.
     The PMip SIP Development Guideline recommends that both receptor and
dispersion models be used to complete the source apportionment.  These two
modeling techniques use different inputs and calculation techniques and
generally produce different results.   EPA's Protocol for Reconciling
Differences Among Receptor and Dispersion Models was used to reconcile these
differences.  The results of the model reconciliation analysis are
summarized in Table 1 .   This procedure ensured that the major contributors
                                     ill

-------
                                      Table 1
                               Final Modeling Results
                    Hayden PM^Q Modeling - Model Reconciliation
                                      All Days
24-hr Impact (ug/m^)
Garfield
Source Group
Auto Exhaust
Roadway Dust
Tailings
Smelter Building
Slag Dump
Copper Ore
Lime Handling
Gypsum
Secondary SO^
Other*
Background
All Sources
Total Measured
CMB
0.
4.
5.
Fug. 0.
0.
73.
0.
1.
4.
13.
—
104.
23
69
29
86
06
75
34
63
11
44

sT
+_
+_
£
£
£
£
£
£
£
+


0.12
2.46
3.60
0.66
0.06
14.73
0.30
0.48
0.43
5.03


ISC
0.42
13.46
0.01
1.02
1.08
61.47
1.06
1.88
—
4.82
16.00
101.22
+ 0
+ 4
+ 0
+ 0
+ 0
+ 18
+ 0
+ o

£ 1


.13
.04
.00
.31
.32
.44
.32
.56

.45


Jail
CMB
0.19
20.71
0
1.04
0.03
27.72
0.57
2.94
4.03
2.29
—
59.49
104.51
+ 0.14
+ 4.69
+ 0.00
+ 0.66
+ 0.03
+ 7.51
+ 0.57
+ 0.71
+ 0.58
±. 1.66


59.49
ISC
0.39
23.21
0
1.11
1.43
12.10
1.65
2.39
—
2.30
16.00
60.58

+ 0.12
+ 6.96
-I- 0.00
+ 0.33
+ 0.43
+ 3.63
+ 0.50
£0.72

+ 0.69



* ISC: Woodburning, Locomotives, Windblown Dust, Stack
  CMB: Unexplained
                                         iv

-------
to the ambient PM^Q concentrations in the Hayden area were sufficiently
identified.

     The 1986 emissions inventory, adjusted through model reconciliation,
together with one full year of meteorological data were used as input to the
ISCST model to estimate the 1986 base year design concentrations both for
the annual arithmetic mean (AAM) and the 24-hour maximums^  The 1986 PM^g
emissions inventory used to determine the design concentrations is given in
Table 2.  Based on criteria given in the Guideline Document and EPA's
Guideline on Air Quality Models (Revised), design concentrations of 618
ug/mj (24-hour) and 103 ug/mj (AAM) were determined and provided the basis
for control strategy development.  The model results identified the major
PMjQ contributors and their relative air quality impacts.  Control options
were compiled and evaluated for each of the major sources and selected
options formulated into control strategies.  The impacts of possible
operation of the second smelter at Hayden (not operated since 1980) and an
estimated 20 percent increase in activity levels of population dependent
sources, were evaluated, as were the effects of the selected control
strategy for the year 1990.  The Kennecott smelter "banked" emissions were
used in the assessment of 1990 ambient PMjg concentrations in order to
address the effect of the possible start-up of the smelter.  Additional
growth between 1990 and 1997 is expected to occur and therefore, the 1997
emissions were increased accordingly.

     The model results showed that the application of controls to major PM^Q
source groups in the Hayden area would result in attainment of the PM^g
NAAQS by 1990, and retain attainment status through 1997.  Emission
reductions would be achieved by the following control strategies:

     (1)  control of copper ore unloading, crushing and conveying systems;

     (2)  paving the unpaved roads in Hayden and placing curbs on the
          heavily travelled paved roads;

     (3)  control locomotive exhaust emissions by modifying engine
          operation;

     (4)  control of gypsum sources by enclosure and use of fabric filters.

     Figures 1 and 2 show the 24-hour and the AAM design values as well as
projected air quality reflecting the impact of growth and implementation of
the control strategy.

     Concentrations of potentially carcinogenic air pollutants were
determined from XRF analysis of the dichotomous filters at the two Hayden
area monitoring sites, and a screening level estimate of cancer risk in
Hayden was conducted.  The calculations used Unit Risk Factors developed by
the California Department of Health Services and EPA and showed total excess
cancer risk to be approximately 1.5 in 10,000 people assuming exposure to
measured ambient concentrations over a 70 year lifetime.  Five non-
carcinogenic elements considered to be toxic were also identified.
Pyrometallurgical sources at the smelter produced most of the toxic

-------
                                 Table 2

                  Summary of 1986 PM^g Emission Inventory
                               for  Hayden,  AZ
                                                        Annual-Average
                                       Source            Emission Rate
Source Category                         Type3               (ton/yr)
Automobile Exhaust
(including brake and tire wear)
Roadway Dust
Woodburning Stoves
Copper Ore Tailings
Locomotive Exhaust
Windblown Dust
ASARCO Smelter Stack
ASARCO Smelter Building Fugitives
ASARCO Slag Dump-
Copper Ore
Ore Crusher
Lime Handling
Gypsum
Total
A

A
A
A
A
A
P
P
P
P
P
P
P

2.0

35.4
0.3
84.9
5.3
2.3
87o5
17.7
40.6
15.3
59.8
13.2
13.2
377.5
a A = area source, P = point source
                                      vi

-------
   700 T
   600 --
O 500
z
O
m
z
-i

>
o


£"

400 ••
   300 ••
   200 •-
m
z
o
z
m
m
2
2
O
l
CO
O
rn
z
o
m
   100 ••
                                       PROJECTED 24-HOUR CONCENTRATIONS

                                              FOR HAYDEN, ARIZONA
                            618
                               138
                                                                            147
1986   1987   1988   1989   1990
                                     1991   1992
                                        YEAR
                                                                                   o
                                                                                   c
                                                                                   V
                                                                                   m
                                               1993   1994   1995   1996   1997

-------
   110 T
   100 - -
   90  -•
o  80
o


I  70
2)

d
o
   50
•s.

3
 u

-  40
   30  ••




   20  •-




   10  -•
 PROJECTED ANNUAL ARITHMATIC


    MEAN CONCENTRATIONS


     FOR HAYDEN, ARIZONA
     1986   1987   1988   1989   1990
1992   1993   1994   1995   1996   1997
o
c
ai
m

10

-------
compounds.  The PM^Q control strategy, because it did not include further
emission reduction from these sources, indicated only minor reductions in
these pollutant concentrations.

-------
1.0  INTRODUCTION

1. 1  Purpose

     This report describes the use of EPA's PMip SIP Development Guideline
Document in the preparation of an example State Implementation Plan (SIP)
for PM^Q, i.e.,  particulate matter with mass mean aerodynamic diameter less
than 10 micrometers.   The example non-attainment area selected was Hayden,
Arizona, a small smelter town located approximately 70 miles southeast of
Phoenix.  The PMm SIP Development Guideline Document, hereafter referred to
as the Guideline Document, emphasizes the use of receptor models for
dispersion model refinement; a Chemical Mass Balance (CMB) type of receptor
model was used in this example.  This report also discusses the development
of PM^o emission inventories and ambient air quality data and provides an
overview of requirements for PMio control strategies.

1 .2  Background

     The following two paragraphs taken from the Introduction to the
Guideline Document summarize Clean Air Act requirements and review
procedures for a revised particulate matter ambient air quality standard and
the development of the PMjQ standard:

     The 1977 amendments to the Clean Air Act require the Environmental
Protection Agency (EPA) at five year intervals to review and, if
appropriate, revise the criteria on which each national ambient air quality
standard (NAAQS) is based along with the NAAQS themselves.  In response to
these requirements, EPA reviewed the criteria upon which the particulate
matter NAAQS were based along with information on health and welfare effects
that had become  available since the original criteria document was prepared
in 1969.  The Criteria Document was revised accordingly, and reissued on
March 20, 1983,

     After considering the information in the revised criteria document, EPA
revised the NAAQS for particulate matter.  Prior to this action the original
particulate matter NAAQS included the size range of particles collected by
the hi-volume sampler and referred to as total suspended particulates (TSP).
The revised primary and secondary NAAQS apply to particulate matter in a
size range defined by the collection characteristics of a new ambient
reference method that has a 50% collection efficiency (D50) at 10
micrometers.  The material collected by the reference method is nominally
below 10 micrometers and is referred to as
     The effective date of the revised NAAQS was July 31,  1987.  The Clean
Air Act requires that states submit revised State Implementation Plans
(SIPs) within 9 months of the effective date showing control strategies that
demonstrate attainment of the new NAAQS.  Based on the proposed NAAQS the
Hayden, Arizona area has been classified as non-attainment for the annual
and 24-hour average PM^Q NAAQS.
                                    1.1

-------
1 .3  Study Area Definition

     The Guideline Document states that the TSP areas of non-attainment may
in some instances differ from PM^Q non-attainment, and suggests that
consideration be given to analyzing the representativeness of the monitoring
stations, and to spatial interpolation of air quality data.  Dispersion
modeling to define the area boundaries should also be considered.
     For Hayden it was determined that the PM^o non-attainment area should
remain consistent with the TSP area designation, which the Code of Federal
Register (40CFR81.303) lists as the entire 36 square mile township (T5S,
R15E), which includes the towns of Hayden and Winkleman, the two copper
smelters, and the copper tailings pile.  Figure 1.1 shows a map of this
township.  For both TSP and PM^g the actual non-attainment area is probably
confined to the towns and the areas of smelter related activities.

1 .4  General SIP Approach

     The basic approach in developing and implementing PM^Q attainment will
be almost the same as for TSP.  As explained in the Guideline Document, that
will be to (1) examine air quality data, (2) inventory the sources
contributing to the problem, (3) determine the areas where air quality needs
improvement with the aid of dispersion models, (4) determine the degree of
improvement necessary, (5) develop a strategy to reduce emissions sufficient
to bring about attainment, (6) implement the strategy, and (7) take the
steps necessary to ensure that NAAQS are not violated in the future.  The
significant changes from the TSP approach are that the emissions inventories
and control strategies must be specific for PM^g, and that receptor modeling
techniques be used to augment the dispersion models.  Dispersion models
reconciled with results of receptor modeling can more accurately define
source-receptor relationships and thus better evaluate control alternatives
and the impact_of growth.

     The procedures described above were used for the Hayden SIP.  Attention
was also given to the sources and impacts of potentially Toxic Air
Contaminants (TAG).

1.5  Report Contents

     Section 2 of this example SIP describes the air quality data bases used
to determine the non-attainment status, and describes the development of an
ambient speciation profile for receptor modeling*  It also describes the
receptor model source speciation data base and the mass PMjQ data base used
for preliminary dispersion modeling.  Section 2 also defines the meteorology
of the area and the meteorological input to the dispersion model.

     Section 3 demonstrates the use of measured size specific and speciated
source profile data in receptor model Chemical Mass Balance (CMB) equations.
It also describes the validation procedures involving statistical comparison
techniques with measured size specific and speciated ambient air quality
data.
                                     1.2

-------
                Hayden,  Arizona Attainment Area
                Township 5 South,  Range 15 East
(sections 9 thru 12,  13  thru 16, and 21 thru 24,  subdivided for
               Receptor  and Dispersion Modeling)

                              1.3

-------
     Section 4 describes the receptor and dispersion model reconciliation
procedures and the resultant emissions inventory reflecting the
reconciliations.

     Section 5 shows the base year modeling and development of 24 hour and
annual maximum design concentrations from which to base control strategies.

     Section 6 describes development and modeling of source specific control
strategy options and projected air quality for 1990 and 1997 using growth
estimates and control strategies.

     Section 7 describes the control strategy implementation methodology and
schedule.

     Section 8 defines the risk to the population of toxic pollutants, the
sources of these pollutants, and the impact of the PM}Q control strategy.

     The appendices contain the completed report prepared by NBA Inc. for
the CMB analysis and receptor modeling, and contain emission factors and
emission calculations for PM   and toxics.
                                    1.4

-------
2.0  DATA BASE DEVELOPMENT

     The Guideline Document lists the combined use of receptor and
dispersion models as the preferred method of establishing PM^g source-
receptor relationships.   In a discussion of receptor models it describes
the Chemical Mass Balance (CMB) method as the most advanced.

     To develop the scope of the needed dispersion and CMB model data base
first requires an overview of the existing PM^Q ambient concentrations and
the major contributing sources.  The more detailed Chemical Mass Balance
requires speciated data for both the ambient air and the emission sources,
and can be further enhanced by fractionated sampling data such as that
obtained with a dichotomous monitor.  The Guideline Document recommends that
at least 80 percent of the ambient PM^o ^e identified by source
characterization.  The dispersion modeling requires coordinate locations of
all major sources and the ambient monitors, a general estimate of the mass
PMjo emissions from each source (emissions inventory), and meteorological
data consisting of wind speed and direction frequencies and mixing height on
specific days and on an annual basis.

     The historical ambient air monitoring data collected in the Hayden area
determined the non-attainment status and the classification of Hayden as a
Group I area, (described in Section 2 of the Guideline Document as an area
shown to have a high probability of non-attainment).  Data from a size
selective inlet (SSI) PM^g monitor and two dichotomous samplers temporarily
installed to obtain speciated and size specific air quality data for
receptor modeling added to the air quality data base.  Source data was
obtained from a source speciation test program and a preliminary PM^Q mass
emissions inventory.  Meteorological data were obtained from the ASARCO
meteorological station in downtown Hayden and other local meteorological
stations, and also from the Tucson airport.  Speciation data for sources not
tested were determined from speciation libraries.  The following subsections
describe the ambient characterization, the source characterization, and the
meteorological data base developed for the Hayden SIP.

2.1  Air Quality

2.1.1  Existing Air Quality Data

     The Arizona Department of Health Services (ADHS) has operated a SSI
PM^Q sampler at the Hayden Jail Monitoring Site since February, 1985, and
has operated a TSP monitor at that same location for preceding years.  Table
2.1 shows approximately 2 years of available PM^Q data and  four years of TSP
data.  The monitoring results show exceedances of the proposed PMiQ NAAQS of
50 ug/m3 (annual arithmetic mean) and the 150 ug/m3 (24-hour average), and
also exceedances of the TSP annual geometric mean standard  of 75 ug/m3 and
24-hour standard of 260 ug/m3.

     Chapter 2 of the Guideline Document describes uncertainties in data
measured with the Sierra Anderson SA321A PM^o monitors, such as the monitor
used at the Hayden Jail site, and the use of a 20 percent "gray zone" around
the standard.  Even allowing for the 20 percent "gray zone" the Jail

                                    2.1

-------
                                 Table 2.1

               Measured Air Quality at the Hayden Jail Site*
  Pollutants,
Averaging Times,
& No. of Samples              1983        1984         1985        1986
     (ug/m3)
   Annual Arithmetic Mean       —         —            68          80
   Maximum 24-hour              —         —           157         243
   2nd High 24-hour             —         —           146         151
   No. of Samples               —         —            57          60

TSP (ug/m3)
   Annual Arithmetic Mean       98         122          123         158
   Maximum 24-hour             337         514          378         318
   2nd High 24-hour            294         301          271         272
   No. of Samples               53          58           58          61
*Does not include dichotomous sampler data collected December 1986 through
 January 1987.
                                    2.2

-------
measurements still show exceedances of the NAAQS but are within the "gray
zone" for the 24-hour second high values.  Using information from the EPA
Procedures for Estimating Probability of Nonattainment of a PMm NAAQS Using
Total Suspended Particulate or PMm Data, referenced in the Guideline
Document, and the 0.47 TSP to PM^g adaptation factor in the Guideline
Document, the measured TSP data also indicates exceedances of the NAAQS for
PM10.

2.1.2  Special PM^g and Toxic Air Contaminant Monitoring for SIP

     Dichotomous monitors using Teflon filters to aid in speciation analysis
collected size specific air quality data at the Hayden Jail site and at the
Garfield Street site in Hayden from the beginning of December 1986 through
January 1987.  The two months of speciated data were subsequently used in
the CMB model and compared with the dispersion model results.  Sampling for
a full year period would have allowed assessment of the impact of seasonal
sources and seasonal meteorology.

     Both the fine and coarse fraction dichotomous sampler filters from each
location were desiccated and weighed to determine total mass and then
examined with X-Ray Fluorescence (XRF) for speciation analysis.  The
speciation analysis and particle coarse and fine size data provided input to
the CMB determinations.  For mass PM^g concentrations (determined from the
sum of the fine and coarse dichotomous filter concentrations), the 12
samples collected at the Hayden Jail and the 13 samples collected at the
Garfield site showed arithmetic means of 105 ug/m^ and 59 ug/m^,
respectively, indicating that one year of sampling would result in annual
standard exceedances.  Two of the 25 samples collected exceeded the 150
ug/m^ 24-hour standard.

2.2  Receptor Modeling Data Base

     The type of receptor model selected determines the type of data base
required.  Chapter 4 of the Guideline Document discusses the considerations
in model selection and the data bases needed.  It describes the air
monitoring and sample collection data base best suited for use with the
Chemical Mass Balance type receptor model.  NBA, Inc« conducted the sample
collection and analysis program and the receptor modeling for Hayden.  NEA's
report "Source Apportionment of Suspended Particles and Toxic Elements in
Hayden, Arizona," hereafter referred to as the NEA Report, is paraphrased in
this report and contained in its entirety as Appendix A.

2.2.1  Receptor Model Selection

     The Guideline Document lists the several factors affecting the choice
of receptor models and states that Chemical Mass Balance is considered the
most advanced of the receptor methods.  Review of these factors relating to
the conditions at Hayden resulted in selection of the CMB approach to best
establish source receptor relationships.
                                    2.3

-------
2.2.2  Ambient Profile Development

     -  Ambient Monitoring Sites

     Two ambient monitoring sites separated by about one half mile were used
to collect ambient particulate samples.  A dichotomous sampler was located
at the Hayden jail site southeast of the Kennecott and Canyon Road
intersection, and the second dichotomous sampler was located at the Hayden
City Maintenance Yard on Garfield Avenue a few hundred yards uphill from the
ore crusher building.  The Hayden Jail site was equipped with a TSP and SSI
     monitor in addition to the dichotomous monitor.

     - Sampling Schedule

     All samples collected were 24-hour, midnight to midnight samples.
Sampling started on November 28, 1986, and continued through January 31,
1987, at both sites.  Dichotomous samples were collected every third day at
both sites from November 28, 1986, to December 7, 1986, and on an every day
schedule from January 27, 1987, through January 31, 1987, the latter during
the period when source samples were collected at the smelter.  From December
18, 1986, through January 12, 1987, samples were collected at the Hayden
Jail site on a schedule of two days on and two days off, while samples were
collected every day for three days at the Garfield site with only one day
off.  In all, 114 fine- and coarse-cut samples were collected at the Hayden
Jail and Garfield sites.  The short overall sampling period prevented
assessment of seasonal influence.

     -  Ambient Filter Selection

     Filters selected for chemical analysis and source apportionment
represented worst case dichot PM^Q days, as well as days representing
conditions that might enhance the impacts of specific sources.  The
selection criteria used in order of importance were as follows:

          - High PM10 Values
          - Predominant Wind Direction
          - Wind Speed
          - Fine to Coarse Ratios
          - Plant Operations
          - Days of Week
          - Unusual Events

     The days and filters selected, their key features, and selection
criteria are summarized in Table 2.2.  The wind speed and direction on the
fifteen days selected were grouped into five regimes based on wind speed and
direction, as indicated in Table 2.3.  The number of days selected from each
regime is generally typical of their historical frequency of occurrence.

     The average wind speeds listed are twenty-four hour averages of five
monitoring sites operated by the ASARCO smelter.  Although the average wind
speed on the highest wind speed days (Regime 1) was less than 6.3 mph, short
term average wind speeds to 24 mph were recorded, and wind speeds in excess

                                    2.4

-------
                             Table 2.2
                    SUMMARY OF KEY FEATURES
FOR SAMPLING DAYS SELECTED FOR ANALYSIS AND SOURCE APPORTIONMENT
Date/Day
U/22/86
Saturday
12/1/86
Monday
12/4/86
Thursday
12/9/86
Tuesday
12/18/86
Thursday
12/24/86
Wednesday
12/28/86
Sunday
12/29/86
Monday
Site
1
2
1
2
1
2
1
1
2
1
2
2
2
Filter
Particle
P0591
P0590
P0595
P0594
P0597
P0596
P0603
P0602
P0609
P0608
P0605
P0604
P0619
P0618
P0675
P0674
P0679
P0678
P06B7
P0686
P0689
P0688
P0701
P0700
P0703
P0702
ID
Size
F
C
F
C
F
C
F
C
F
C
F
C
P
C
F
C
F
C
F
C
F
C
F
C
F
C
Wind Speed Wind P/C
(mph) •' Direction* Ratio
2.4 261 0.254
0.288
• 2.1 106 0.319
0.211
3.2 121 0.337
0.232
1.4 229 0.265
1.3 156 0.739
0.331
0.9 211 0.405
0.452
3.9 76 0.112
6.2 79 0.065
Dichot PMi(y
(M8/*J)
70.85
72.24
99.27
160.20
126.91
147.46
55.72
25.48
46.93
45.40
42.99
109.65
235.69
Comments & Selection
Criteria
Furnace down 0200-0530
Moderate PMiQ values and
wind from the west
2nd High PM10 (Site 1 & 2)
Furnace down 0630-0830
Highest PMlo (Site I)
All systems down 0130-143C
Highest F/C ratio
Wind gusts from the west
Holiday
High F/C ratios
Furnace down 1130-1530
Highest PMlo (Site 2)
Lowest F/C ratio
Furnace down 0900-1200

-------
  Table2,2
-Continued-
Date/Day
1/4/87
Sunday
1/7/87
Wednesday
1/11/87
Sunday
1/12/87
Monday
1/27/87
Tuesday
1/28/87
Wednesday
1/30/87
Friday
*Wind
east
Site
1
1
2
1
2
2
1
2
1
2
1
2
Filter
Particle
P0715
P0714
P0751
P0750
P0755
P07S4
P0761
P0760
P0763
P0762
P0767
P0768
P0783
P0784
P0779
P0780
P0785
P0786
P07B1
P0782
P0791
P0792
P0787
P0788
ID
Size
F
C
F
C
F
C
F
C
F
C
F
C
F
C
G
e
F
C
F
C
F
C
F
C
direction based upon 360
90°, due south 180°, and
Wind Speed
(mph)
4.8
0.8
6.6
5.0
4.5
.1.6
2.4
Wind
Direction*
116
181
81
86
132
161
124
0 circle. Winds out of due
winds out of due west 270°
F/C
Ratio
0.396
0.502
0.187
0.218
0.056
0.165
0.232
0.394
0.314
0.330
0.180
0.086
geographic
»
Dlchot PMlo
(ug/m1) .
48.04
34.09
62.19
34.36
126.69.
111.06
56.87
31.91
78.03
82.87
71.47
128.47
north would
Comments & Selection
Criteria

All systems down 0630-1830
Stagnant conditions
Lowest average wind speed
Furnace down 2000-2230
Windy, low F/C ratio
Furnace down 0930-1200
and 1730-1800

Furnace down 2130-2400
Rain, F/C ratios low and
PM}Q values relatively
high
measure 0°, due

-------
                Table 2.3

DAIS INCLUDED IN METEOROLOGICAL REGIMES
      AND REGIME CHARACTERISTICS
No.
1
2
3
4
5
Description
High vind speed from the East
Low vind speed from the South
Moderate vind speed from the
Southeast
Moderate vind speed from the
West
Moderate vind speed from the
East
Speed/Direc tion
(>5 mph, 79°-86°)
(<1.6 mph, 156*-211°)
(>2,<5, 106°-132°)
(2.4, 261°)
(3.9, 76°)
Dates Included
12/29/86, 1/11/87
1/12/87
12/9/86, 12/18/86
12/24/86, 1/7/87
1/28/87
12/1/86, 12/4/86
1/4/87, 1/27/87
1/30/87
11/22/86
12/28/86
                2.7

-------
of 15 mph accounted for a  significant portion of the sampling period.

2.2.3  Source Profile Development

     - Sample Collection

     Composition profiles  representing emissions from sources expected to
significantly impact the ambient monitoring sites were developed from
literature values and direct characterization by measurement of emissions
from Hayden area sources.  Samples representing area fugitive dust and
process fugitive emissions were collected and characterized by procedures
similar to those used to characterize the ambient aerosol.  Process
fugitives, such as those from the ASARCO smelter activities were
characterized by sample collection on-site using a sampler developed by NBA,
Inc.  This sampler incorporates the dichotomous sampler principle and uses a
dilution chamber to allow  isokinetic sample collection.  The test team
collected samples of emissions from the secondary converter hoods, the
matter tapping and slag skimming of the flash furnace, the slag pouring
operation, and the ore crusher.  The NEA Report (Appendix A) describes the
testing of each source.

     Non-process fugitive  sources were characterized by collection of bulk
samples which were then resuspended and collected onto dichotomous filters
in the laboratory.  This method was used for characterization of unpaved
roads, areas of disturbed  soil, ore concentrate charged to the furnace,
settled dust in the ore crusher area, and the smelter tailings pile.  The
compositing of several samples prior to resuspension allowed more
comprehensive characterization.

     -  Selection of Filter Samples for Analysis

     Filter samples collected from sources of fugitive emissions are listed
in Tables 2.4 and 2.5.  Only nineteen of the process fugitive emission
filters were selected for  analysis as indicated by asterisks in Table 2.4.
The criteria for selection included optimum deposit mass, uniformity of
deposit, duration of sampling, availability of other filters representing
the same process and resource limitations.  All of the resuspension filters
listed in Table 2.5 were analyzed.

2.3  Dispersion Modeling Data Base

2.3.1 Model Selection

     The first step in establishing a data base for dispersion modeling
analysis is to determine the appropriate model to be used.  The Guideline on
Air Quality Models, US EPA, July 1986 should be reviewed to determine the
appropriate model, and the selection of the model should be reviewed with
the Regional Meteorologist or other EPA personnel.  Once selected, the input
requirements of the model  dictate the type of data needed for the analysis.

     For Hayden, an EPA approved model, Industrial Source Complex Short-Term
(ISCST) Version 6.0 was selected for evaluating PM^g impacts.  This model

                                    2.8

-------
                                              Table  2.4
                                     SUMMARY OF SAMPLING DATA
                                       FOR PROCESS EMISSIONS
                           COLLECTED AT THE ASARCO-1IAYDEN COPPER SMELTER
Filter
10
*P0949
•P0950
P0951
P0932
•P09S3
•P09S4
*P095S
*P0956
P09S7
P0958
*P096l
•P0962
•P096S
•P0966
»P0967
*P0968
•P0969
»PG970
•P097I
•P0972
•P0971
•P0974
*P097i
P0976
P0977
P0978
Run
1
I
1
2
2
3
3
4
4
i
3
6
6
2
2
3
3
4
4
1
I
2
2
1
1
2
2
Filter
Source Description Type
Slag ekla; flesh furnace
Slag eklJii dealt furnace
Katie fc alag tap; (lath (urn.
Matte i gleg tepi flaeh (urn.
Hate* cep| (leeh furnace
Matt* tap; flaah furnace
Matte, tap; flaah furnace
Matte taps (leah furnace ,
Hatte t elag tapi flesh (urn.
Matte ft aleg tapi flaah (urn.
Slag »kl»; flaah furnace
Slag ikt>i flash furnace
Secondary converter ducting
Secondary converter duct log
Secondary converter ducting
Secondary converter ducting
Secondary converter ducting
Secondary converter ducting
Slag pour
Slag pour
Slag pour
Slag pour
Ore crusher building
Ore crueher building
Oru cr untie r building
Ore crueher building
T
T
T
t
T <
T
T
T
T
T
T
T
t
T
T
T
T
T
T
T
T
t
T
T
T
T
Particle
Slse (|i)
'• <2.5
>2.S and 2.i and <10
<2.J
>2.i and <10
<2.S.
>2.S and <10
<2.S
>2.S and <10
<2.i
>2.J and <10
<2.S
>2.S and 2.i and <10
<2.5
>2.S and «iti
<2.i
>2.) and 2.S and <10
<2.i
>2.3 and <10
<2.i
>2.5 and 
8372
870
1934
134
10232
682
14379
994
30377
2071
2934
334
17381
924
IS072
1311
3407
335
1102
51
1232
69
412
3366
Coemeote

A
I
I
A

C
C
C
D
0
B
1
Comment Key:
  A.  Combination of slag skimming and matte tap
  B.  Two consecutive matte taps during sample time
  C.  See Table   for converter activity
  D.  Two slag pours per filter pair; coarse filter collimated
  E.  All three crushers running
  F.  37 mm Teflon membrane
  ^Filters selected for XRF analysis

-------
                                     Table 2.5




                     Summary of ASARCO-Hayden Resuspension Data
Description
Charged ore
mixture
Composited
Smelter Yard and
Road Dust
Dust from Ore
Crusher
Sample
Type*
dichot fine
dichot coarse
dichot fine
dichot coarse
dichot fine
dichot coarse
Filter
ID
P1029
P1030
P1031
P1032
P1033
P1034
Final Net
Deposit (ug)
182
2361
200
2489
199
2555
Ratio of Net Intermediate
Fine to Net Coarse
0.076
(1:13)
0.076
(1:13)
0.042
(1:24)
*Fine is <2.5 urn, Coarse is 2.5 urn - 10 um.
                                        2.10

-------
was selected because of its ability to:

          model point, area and volume sources;

          calculate combined impacts for selected source groups;

          vary windblown dust source emissions as a function of wind speed;

          calculate short-term (1 to 24-hours) and long-term (annual)
          concentration using hourly-sequential meteorological data;

     calculate concentrations at designated receptor locations (i.e.
     ambient. monitoring stations).

     This model allows for the evaluation of PMjQ impacts using an emission
inventory of the study area (containing both area and point sources) and
hourly meteorological data.  The ability of the model to group sources aids
in the comparison and reconciliation of modeled results with the results
from the CMB receptor modeling analysis.

2.3.2 Emission Inventory

     An emission inventory of the PM^g sources within the study area must be
compiled for input into the dispersion model.  The inventory should contain
source locations as well as PM^g emission rates and release parameters.  The
emission rates are generally based on published or site-specific emission
factors and known or estimated activity levels.

     Ideally, two types of inventories would be created; day-specific and
annual-average.  The former would be used in the receptor/dispersion
reconciliation analysis while the latter would be used for control strategy
development.  Typically, however, day-specific activity information is not
available and only an estimated annual-average inventory is available for
modeling purposes.  For Hay den, the annual-average emission inventory was
used for both model reconciliation and control strategy development
modeling.
     Annual-average PM^g emissions in the Hayden area were estimated based
on 1986 activity levels and assigned to the 13 general categories given in
Table 2.6.  The total emissions from the 136 area sources and 10 point
sources contained within the study area are also presented in Table 2.6. The
area sources were the tailings pile south and west of Hayden, the railroad
activity south of Hayden near the ore dump area, the windblown dust from
disturbed soil areas in the smelter and town areas, and the tailings and
fugitive emissions from traffic on paved and unpaved streets and roads.
Figure 2.1 shows the point sources and monitor locations in Hayden.  The
actual inventory, containing the source locations, emission factors,
emission rates, release parameters and other descriptive information is
contained in Appendix B.
                                    2.11

-------
                                                  ribwrtc '4.
•  SOURCES
1  LIME HANDLING
2  LIME HANDLING
3  ORE STORAGE
4  CONVEYOR TRANSFER
5  CRUSHER BUILDING
6  ORE DUMP
7  ASARCO STACK
8  SMELTER BLDG. FUG.
9  SLAG DUMP
I0  GYPSUM LOADING
B  MONITORS
J  JAIL
G  GARFIELD
 HAYDEN POINT SOURCES
USED IN DISPERSION MODEL
                     u/-
                                         fit    ^

 v •  	 ~_j	 _;


-------
                                Table 2.6
                  Summary of 1986 PM^g Emission Inventory
                              for Hay den, AZ
                                                        Annual-Average
                                       Source            Emission Rate
Source Category                         Type3               (ton/yr)
Automobile Exhaust
(including brake and tire wear)
Roadway Dust
Woodburning Stoves
Copper Ore Tailings
Locomotive Exhaust
Windblown Dust
ASARCO Smelter Stack
ASARCO Smelter Building Fugitives
ASARCO Slag Dump
Copper Ore
Ore Crusher
Lime Handling
Gypsum
Total
A

A
A
A
A
A
P
P
P
P
P
P
P

2.0

35.4
0.3
84.9
5.3
2.3
87.5
17.7
40.6
15.3
59.8
13.2
13.2
377.5
a A = area source, P = point source
                                    2.13

-------
2.3.3 Meteorological Data

     Meteorological data representative of the study area must be selected
for use in the dispersion modeling analysis.  The PMm SIP Development
Guideline states that either 5 years of off-site or 1 year of site-specific
meteorological data should be used in the development of a PM^Q SIP.  The
meteorological parameters generally required for predicting short-term and
annual average impacts with EPA guideline models are wind speed, wind
direction, ambient temperature, atmospheric stability and mixing height.

     At Hayden, ASARCO routinely collects and processes wind direction, wind
speed, and temperature data at monitoring locations near their smelter.  The
closest off-site station was Tucson airport located approximately 100 miles
southwest of Hayden.  The on-site meteorological data were selected for the
dispersion/receptor modeling reconciliation and for control strategy
development because they are more representative of the wind and temperature
fields in the study area than Tucson data.  The on-site wind data was
supplemented with cloud cover and mixing height data from Tucson to generate
the atmospheric stability and mixing height profile.

     For the comparison and reconciliation of receptor and dispersion
modeling results, meteorological data from the days selected for CMB
analysis were used.  These data were collected in November and December
1986, and January 1987.

     The data used for control strategy development were from one year of
ASARCO hourly sequential data collected in 1983.  Cloud cover and mixing
heights from the Tucson airport for 1983 were obtained and combined with the
ASARCO wind and temperature data to generate an annual sequential
meteorological data base.  These data were used to calculate 24-hour and
annual-average PM^g concentrations to determine the level of controls needed
to attain and maintain the NAAQS.
                                    2.14

-------
3.0  CMB RESULTS AND MODEL VALIDATION

     The CMB provides a source contribution estimate (SCE) and associated
uncertainty (STDERR) for each source category that is chosen beforehand for
consideration with the model.  The model produces this estimate by making an
effective variance weighted least squares "fit" which considers the chemical
composition of the ambient sample and the composition of the sources.  It
estimates what amounts of each source (the SCE's) will collectively best
"explain" the chemistry of the ambient sample.  Generally, it is desirable
to obtain the best "fit" possible given the physical constraints of the
models assumptions and also to obtain SCE's with low STDERR' s.

     The initial steps for the model are to organize its data and perform
the chemical mass balance calculations, followed by validation procedures to
evaluate and adjust the results.  The first portion of this section
describes the data collection and analytical results such as measured mass,
ambient and source particulate chemistry, and the results of the model
calculations .

     The second portion of this section describes the validation procedures
and the final model results.

3.1  Analytical and CMB Results

3.1.1  Measured Particulate Mass Concentrations

     The coarse and fine dichotomous sampler results and the total PM^Q
concentrations were used as part of the filter selection criteria for CMB
analysis.  The results also define the ambient PM^Q concentrations relative
to the NAAQS.  The Hayden dichotomous monitors operated during an
approximate two month period from November 22, 1986 through January 30,
1987;  the Hayden Jail site monitor collected 30 samples during that period,
and the Garfieid monitor collected 24 samples.  Each sample consisted of a
PM2.5 fraction and a PM2.5-10 fraction.  On two occasions the monitors
collected two samples during a 24-hour period to assess diurnal influences.
The NBA Report lists the results for each sampling period*  Table 3.1 lists
the mean concentrations and ranges of all samples collected, and the mean of
the 12 selected days for the -Hayden Jail site on the 13 selected days for
the Garfieid site.

     The highest PM^g value of 236 ug/m^ was measured at the Garfieid site
on December 29, 1985.  Samples were not collected on this day at the Hayden
Jail site.  The second highest PM^g concentration for both sites, 160 ug/m^
for Garfieid and 99 ug/m^ for the Hayden Jail, were measured on December 1,
1986.  The third highest PM^g concentration at the Garfieid site of 147
ug/m3 occurred on December 4, 1986, which was the same day as the highest
Hayden Jail reading of 127 ug/rn-^.  High measured concentrations at the
Hayden Jail site were consistently lower than at the Garfieid site.
     The overall average PM^g value at the Hayden Jail site of 47 ug/m^ was
also substantially lower than the 73 ug/nP measured at the Garfieid site.
                                    3.1

-------
                                 Table 3.1
              COMPARISON OF MEAN SUSPENDED PARTICULATE MASS
                        AT HAYDEN SAMPLING SITES
 Site                                                               	
  No.      Description        Fine         Coarse        PM10       F/C Ratio*
          *•
   1      Hayden Jail      12   ±  6     35   ± 22    47   ±  28   0.43 ± 0.26
            Range (n-30)    5.8 - 32      6.2 - 95    12.0 - 127   0.18 ± 0.7


   2      Garfield .        13   ±  5     60   ±  46   73   ±  49   0.35 £ 0,79
            Range (n=24)    6.8 - 28.0    4.4 - 221   11.6 - 236   0.05 ± 1.61
  Days
Selected
          Hayden Jail         14.6          44.9         59.5          0.32
            (n-12)

          Garfield            15.0          89.6        104.5          0.17
            (n-13)
   *Average fine to coarse ratio
                                    3.2

-------
The average fine particle mass was nearly the same at both sites, 12 ug/m3
at the Hayden Jail site and 13 ug/m3 at the Garfield site.

     The days selected for chemical analysis consisted of about half of the
days sampled.  The analysis days selected were biased towards high PM^Q days
as indicated by the selection criteria (Section 2.2.2) and a comparison of
the average mass values listed in Table 3.1 for sampled and selected days.

     The fine to coarse ratios were generally substantially less than 1.0
indicating a preponderance of coarse particles.  These ratios were lowest on
the high wind speed days and days with highest PM^Q values, indicating an
increased coarse particle mass percentage, while the highest fine to coarse
particle ratios were on the lowest wind speed days, indicating a higher
proportion of the mass as small particles.

3.1.2  Ambient Particulate Chemistry

     The average fine and coarse particle elemental compositions are listed
in Table 3.2 and 3.3.  The most abundant element in both size fractions was
Si, which represented about 20 percent of the mass.  The most abundant
coarse particle element at the Hayden Jail site was Ca followed by Fe and
Al, while Fe and Al were the second and third most abundant elements at the
Garfield site.  Sulfur was the second most abundant fine particle at both
sites followed by Al.

     Most of the Si is in the coarse particle size fraction, with average
fine to coarse particle ratios of 0.26 at the Hayden Jail and 0.14 at the
Garfield site.  Other elements mostly in the coarse fraction include Al, K,
Ca, Ti, Mn, Fe, and Cu.  On the other hand, S, Cl, Zn, As, Cd, Sb, and Pb
are more abundant in the fine particle size fraction.

     The concentration of each of the toxic elements of As, Cd, Cr, Ni and
Pb were below 1.0 ug/m3; average Cu concentrations were just above 1 ug/m^.
The Pb concentration is less than many urban airsheds because of the large
contribution from motor vehicle tailpipe emissions in urban airsheds.  The
As, Cd, Cu and Ni are substantially higher than urban airsheds.

3.1.3  Elemental Composition of Source Profiles

     The potential major sources of PMjQ and toxic elements in Hayden were
grouped into the following eight general source categories:

     - Fugitive area dust sources from wind and traffic
     - Ore handling and processing
     - Fugitive emissions from the smelting process
     - Miscellaneous smelter dust
     - Wind-blown dust from ore tailings ponds
     - Stack emissions from the smelting process
     - Motor vehicle emissions
     - Background
                                    3.3

-------
                  Table 3.2
AVERAGE FINE PARTICLE ELEMENTAL COMPOSITION
Species
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Ni
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Y
Zr
Mo
Pd
Ag
Cd
In
Sn
Sb
Ba
La
Hg
Pb
ug/m3
Eayden Jail
0.4858 £ 0.5228
2.5157 £ 2.4963
0.0119 £ 0.0184
- 1.1033 £ 0.4546
0.0238 ± 0.0177
0.2325 ± 0.1823
0.3538 t 0.2983
0.0454 £ 0.0547
0.0013 ± 0.0019
0.0009 £ 0.0012
0.0087 ± 0.0169
0.3728 ± 0.3791
0.0013 ± 0.0014
0.2237 ± 0.1588
0.0800 ± 0.0765
0.0006 ± 0.0007
0.0791 ± 0.0801
0.0113 £ 0.0061
0.0039 ± 0.0018
0.0004 £ 0.0008
0.0012 ± 0.0013
0.0000 ± 0.0003
o-.oooo ± o.ooio
0.0013 t 0.0039
0.0000 ± 0.0006
0.0000 ± 0.0008
0.0054 ± 0.0067
0.0000 ± 0.0013.
0.0000 ± 0.0016
0.0038 ± 0.0132
0.0000 ± 0.0064
0.0000 £ 0.0118
0.0004 ± 0.0004
0.1664 ± 0.1152
Gar field
0.5498 ± 0.4604
2.8617 ± 2.6311
0.0084 £ 0.0137
1.1017 ± 0.4476
0.0507 ± 0.1001
0.2636 ± 0.2064
0.2111 £ 0.1757
0.0468 £ 0.0507
0.0021 £ 0.0024
0.0008 £ 0.0011
0.0116 £ 0.0202
0.3865 £ 0.3716
0.0015 £ 0.0015
0.1679 £ 0.0923
0.0748 £ 0.0432
0.0006 £ 0.0005
0.0963 £ 0.0998
0.0127 £ 0.0098
0.0034 £ 0.0022
0.0003 £ 0.0008
0.0010 £ 0.0012
0.0001 £ 0.0003
0.0000 £ 0.0009
0.0015 £ 0.0035
0.0000 £ 0.0005
0.0000 £ 0.0007
0.0052 £ 0.0069
0.0000 £ 0.0011
0.0000 £ 0.0014
0.0017 £ 0.0062.
0.0000 £ 0.0056
0.0000 £ 0.0102
Percent
Hay den Jail
3.335 t 3.576
17.273 £17.059
0.082 t 0.126
7.575 £ 3.036
0.163 £ 0.120
1.596 £ 1.242
2.429 £ 2.035
0.312 £ 0.375
0.009 £ 0.013
0.006 £ 0.008
0.060 £ 0.116
2.560 £ 2.592
0.009 £ 0.010
1.536 £ 1.080
0.549 £ 0.523
0.004 £ 0.005
0.543 £ 0.547
0.078 £ 0.041
0.026 £ 0.012
0.003 £ 0.005
0.008 £ 0.009
0.000 £ 0.002
0.000 £ 0.007
0.009 £ 0.026
0.000 £ 0.004
0.000 £ 0.006
0.037 £ 0.046
0.000 £ 0.009
0.000 £ 0.011
0.026 £ 0.090
0.000 £ 0.044
0.000 £ 0.081
0.0004 £ 0.0004 i 0.002 £ 0.003
0.1737 £ 0.0834 1.143 £ 0.783
Garfield
3.667 £ 3.050
19.086 £17.452
0.056 £ 0.091
7.348 £ 2.901
0.338 £ 0.667
1.758 £ 1.366
1.408 £ 1.164
0.312 £ 0*337
0.014 £ 0.016
0.006 £ 0.007
0.077 £ 0.135
2.578 £ 2.466
0.010 £ 0.010
1.120 £ 0.606
0.499 £ 0.284
0.004 £ 0.003
0.643 £ 0.663
0.084 £ 0.065
0.023 £ 0.015
0.002 £ 0.005
0.007 £ 0.008 ;
0.001 £ 0.002 !
0.000 £ 0.006
0.010 £ 0.023 i
0.000 £ 0.004 !
0.000 ± 0.005
0.034 £ 0.046
0.000 £ 0.008
0.000 £ 0.009
0.011 £ 0.041
0.000 £ 0.037
0.000 £ 0.068
0.003 £ 0.003
1.159 £ 0.545
                   3.4

-------
                   Table 3.3
AVERAGE COARSE PARTICLE ELEMENTAL COMPOSITION

Species
Al
'•• Si
P
s
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Ni
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Y
Zr
Mo
Pd
Ag
Cd
In
Sn
Sb
Ba
La
Hg
Pb


/
U8/
Hayden Jail
2.1035 ± 1.3704
9.7309 t 2.1813
0.0276 ± 0.0585
0.8378 ± 0.4305
0.0255 ± 0.0198
0.8308 ± 0.5338
3.6350 ± 2.8756
0.2373 ± 0.2005
0.0126 ± 0.0082
0.0000 ± 0.0003
0.0471 ± 0.0413
2.2803 ± 1.5360
0.0015 ± 0.0017
0.9055 ± 0.6046
0.0384 ± 0.0271
0.0003 ± 0.0008
0.0148 ± 0.0127
0.0007 ± 0.0011
OiOQOS ± 0.0016
0.0039 ± 0.0036
0.0142 ± 0.0096
0.0003 ± 0.0005
0.0000 ± 0.0010
0.0142 ± 0.0115
0.0000 ± 0.0005
0.0000 ± 0.0007
0.0000 £ 0.0009
0.0000 ± 0.0011
0.0000 ± 0.0014
0.0000 ± 0.0029
0.0000 ± 0.0055
0.0000 ± 0.0099
0.0000 t 0.0002
0.0342 ± 0.0191


a3
Garfield
4.6364 ± 2.2458
20.7171 ± 2.5052
0.0308 ± 0.0443
1.0524 ± 0.4497.
0.0216 ± 0.0296
2.6364 ± 1.6353
2.0362 ± 1.3292
0.6505 ± 0.5663
0.0280 ± 0.0237
0.0000 ± 0.0005
- 0.1042 ± 0.0971
4.9837 ± 4.0328
0.0036 ± 0.0028
1.0881 ± 0.5799
0.0615 ± 0.0363
0.0005 ± 0.0016
0.0221 ± 0.0215
0.0013 ± 0.0015
0.0003 ± 0.0007
0.0138 ± 0.0108
0.0200 ± 0.0113
0.0033 ± 0.0037
0.0007 ± 0.0019
- 0.0297 ± 0.0170
0.0000 ± 0.0006
0.0000 ± 0.0007
0.0000 ± 0.0010
0.0000 ± 0.0012
0.0000 ± 0.0014
0.0000 ± 0.0031
0.0118 ± 0.0298
0.0066 ± 0.0237
0.0004 ± 0.0009
0.0453 ± 0.0234


Perce
Hayden Jail
4.685 ± 3.017
21.674 ± 4.353
0.062 ± 0.130
1.866 ± 0.941
0.057 ± 0.044
1.850 ± 1.175
8.096 ± 6.354
0.529 ± 0.444
0.028 ± 0.018
0.000 ± 0.001
0.105 ± 0.091
5.079 ± 3.384
0.003 ± 0.004
2.017 ± 1.332
0.086 ± 0.060
0.001 ± 0.002
0.033 ± 0.028
0.002 ± 0.002
0.002 ± 0.004
0.009 ± 0.008
0.032 ± 0.021
0.001 ± 0.001
0.000 ± 0.002
0.032 ± 0.025
0.000 ± 0.001
0.000 ± 0.002
0.000 ± 0.002
0.000 ± 0.002
0.000 ± 0.003
0.000 ± 0.007
0.000 ± 0.012
0.000 ± 0.022
0.000 ± 0.000
0.076 ± 0.042


sat
Garfield
5.182 ± 2.456
23.156 ± 1.579
0.034 ± 0.049
1.176 ± 0.489
0.024 ± 0.033
2.947 ± 1.804
2.276 ± 1.468
0.727 ± 0.629
0.031 ± 0.026
0.000 ± 0.001
0.116 ± 0.108
5.570 ± 4.473
0.004 ± 0.003
1.216 ± 0.637
0.069 ± 0.040
0.001 ± 0.002
0.025 ± 0.024
0.001 ± 0.002
i
0.000 ± 0.001
0.015 ± 0.012
0.022 ± 0.012
0.004 ± 0.004
0.001 ± 0.002
0.033 ± 0.019
0.000 ± 0.001
0.000 ± 0.001
0.000 ± 0.001
0.000 ± 0.001
0.000 ± 0.002
0.000 ± 0.003
0.013 ± 0.033
0.007 ± 0.026
0.000 ± 0.001
0.051 ± 0.026



-------
Representative source profiles were developed through direct source sampling
for the first five categories.  Literature composition source profiles were
used to represent motor vehicle emissions as well as other miscellaneous
sources.  Source profiles were not developed to represent emissions from the
main stack, acid plant, and bell damper because these sources have been
estimated to have minimal impacts on the local airshed.  The background
aerosol was not characterized.

     The NEA Report attached as Appendix A contains tables listing the
source codes, the source composition profiles for the key sources, and the
fine to coarse particle ratios for each source type.  The tables show that
crustal sources are primarily large particles high in Al, Si, Ca and Fe;
plant road dust shows reduced Ca with high copper and sulfur.
Pyrometallurgical smelter fugitives were generally fine particulate
characterized by concentrations of Zn, As, Cd, Sb, and Pb.  Subtilties in
the relative concentrations of these elements distinguished the various
sources within the smelter.  The tables and text describing source
characteristics are contained in the NEA report.

3.1.4  Source Impacts

     Chemical mass balance source apportionment calculations were performed
on fifty fine and coarse particulate filters, representing about one-half of
the sampling days in this study.  Individual source apportionment results
for each filter are presented in the NEA Report.  Table 3.4 shows an example
of the CMB results.  In this example, the filter identification numbers for
the matching fine and coarse particle filters (715, 714), particle size
(coarse), sampling site (Hayden Jail, HD), and sampling date (January 4,
1987) are listed at the top of the page.  The CMB modeling performance
measures are listed just below this information, (R-SQUARE:  .9749, CHI
SQUARE:  1.3403), followed by a listing of the fitted sources and their
source contributions.  In the bottom portion of the table are listed the
measured elemental concentrations, their calculated concentrations, and the
ratio of the calculated to measured concentrations.  Elements actually used
in the fitting process are indicated with asterisks.

     The CMB results presented in the Appendices represent the best model
solution as determined by an -iterative procedure which optimizes the
following model performance parameters:

     -  Source contributions should be positive and greater than their
        uncertainties.  The T-statistic value should be greater than 2.0.

        R-square values should be greater than 0.8.

     -  Reduced chi square should be minimized and generally less than 2.  A
        value greater than 4 indicates that the model has not explained the
        ambient data well.

     -  The calculated to measured concentration ratio for individual
        elements should approach 1.0 within the listed uncertainty.


                                    3.6

-------
RESULTS FOR CMS SITE: HO 715 714
COARSE PARTICULATE FRACTION
SAMPLING DURATION: 24 MRS. WITH START HOUR: 24
R-SQUARE:       .9749
CHI  SQUARE:     1.3403
OF:                8
                                        Table  3.4

                        Sample Source  Apportionment  Results
                                       Hayden  Jail
                                     Coarse Fraction
                                     January 4, 1987
             TEAR: 87  DATE: 0104
     TYPE
UG/M3
5
7
13
16
19
22
28
5095
CTPSU
XN6CT
SSFF1
SSFF2
2NCD3
OUSTO
.000*- .000 .000*- .000
1.945*- .481 5.651*- 1.509
8.833*- 2.067 25.658*- 6.538
.000*- .000 .000*- .000
.176*- .032 .510*- .106
.000*- .000 .000*- .000
22.818*- 2.034 66.280*- 8.917
TOTAL: 23.772*- 1.855 98.098*- 11. 260
HISS COARSE SUSPENDED P ARTICULATE
SPECIES INCL FLC NEAS. UG/M3
1
13
14
15
16
17
19
20
22
23
24
25
26
28
29
30
31
33
34
35
37
38
39
40
42
46
47
48
49
50
51
56
57
80
82
TOTAL
AL •
SI »
P
S *
a
1C •
CA •
TI •
V
CR
NN •
FE •
HI
at
ZN *
GA
AS •
SE
BR •
RB
SR
T
zx
NO
PO
AC
CO
IN
SN
SB
BA
LA
HC
PB •
34.42620*-
1.88920*-
9.02600*-
<
.63590*-
.02030*-
.78350*-
1.72870*-
.17810*-
<
<
.03080*-
1.67830*-
<
.68500*-
.03000*-
<
.00960*-
.00070*-
.00150*-
.00300*-
.00790*-
.00160*-
<
.01110*-
<
<
<
<
<
<
<
<
<
.02610*-
3.46960
.23980
1.01600
.00000
.07620
.00680
.08870
.19420
.04050
.00890
.00000
.00360
.18880
.00020
.07750
.00540
.00000
.00350.
.00030
.00040
.00060
.00100
.00160
.00000
.00200
.00000
.00000
.00000
.00000
.00000
.00000
.00000
.00000
.00000
.00410
PERCENT
100.00000*-
5.48768*-
26.21840*-
«
1.84714*-
.05897*-
2.27588*-
5.02147*-
.51734*-
<
<
.08947*-
4.87507*-
<
1.98976*-
.08714*-
<
.02789*-
.00203*-
.00436*-
.00871*-
.02295*-
.00465*-
<
.03224*-
<
<
<
<
<




.07581 -
14.25297
.88943
3.96132
.03340
.28922
.02063
.34496
.75785
.12868
.02656
.00203
.01381
.73632
.00145
.30149
.01798
.00116
.01055
.00090
.00124
.00195
.00371
.00467
.00930
.00666
.00465
.00668
.00842 *
.01075
.01249
.02643
.05054
.09266
.00087
.01415
CALC. U6/H3
33.77156*-
1.98253*-
7.27240*-
.00487*-
.61063*-
.01752*-
.88214*-
1.87016*-
.25115*-
.00505*-
.00245*-
.02618*-
1.59822*-
.00320*-
.26821*-
.02874*-
.00026*-
.01674*-
.00011*-
.00077*-
.00497*-
.00765*-
.00165*-
.00069*-
.00698*-
.00000*-
.00000*-
.00080*-
.00089*-
.00147*-
.00350*-
.00000*-
.00619*-
.00004*-
.02014*-
1.85531
.16024
.60461
.00578
.04299
.00308
.08102
.12443
.02292
.00194
.00047
.00204
.13348
.00246
.01978
.00361
.00033
.00374
.00046
.00320
.00044
.00056
.00033
.00110
.00081
.00063
.00084
.00113
.00134
.00154
.00355
.00667
.01143
.00031
.00287
RATIO
.98*-
1.05*-
.81*-
.00*-
.96*-
.86*-
1.13*-
1.08*-
1.41*-
.57*-
.00*-
.85*-
.95*-
16.01*-
.39*-
.96*-
.00*-
1.74*-
.15*-
.52*-
1.66*-
.97*-
1.03*-
.00*-
.63*-
.00*-
.00*-
.00*-
.00*-
.00*-
.00*-
.00*-
.00*-
.00*-
.77*-
.11
.16
.11
.00
.13
.33
.16
.14
.35
.62
.00
.12
.13
41.86
.05
.21
.00-
.75
.66
2.14
.36
.14
1.05
.00
.13
.00
.00
.00
.00
.00
.00
.00
.00
.00
.16
TOTAL
AL
SI
P
S
CL
K
CA
TI
V
CR
MN
FE
NI
cu
ZN
GA
AS
SE
BR
RB
SR
Y
ZR
MO
PO
AC
CD
IN
SN
SB
BA
LA
HG
PB
MEASURED AMBIENT MASS (UG/M3):  FINE:   13.6*- 1.4  COARSE:  34.4*-  3.5   TOTAL:   48.0*- 3.8
                                             3.7

-------
     -  The percent elemental mass explained should approach 100 percent
        within the uncertainty.

     -  The degrees of freedom should be maximum, preferably greater than 5.

     -  Source contribution estimates should approximate the measured mass
        concentrations.

In the example illustrated in Table 3.4, 98.098 +_ 11.260 percent of the mass
was explained by the sources indicated.  A reduced chi square value of
1.3403 was obtained with 8 degrees of freedom (DF).  The calculated to
measured elemental ratios were generally equal to 1.0 within their listed
uncertainties for the fitting elements.  The largest deviations from 1.0
were for elements below the analytical detection limit and for abundant
elements like Cu which were omitted from the fitting because of the lack of
a source capable of explaining the excess copper as discussed in the next
section.  The R-square value is 0.9749, and the mean TSTAT value is 4.98
with a low value of 3.74.

     Ore dust from the ore crusher, conveyor belt, etc., was the largest
source of coarse and PM}Q suspended particulate mass at both the Hayden Jail
and Garfield monitoring sites.  Kennecott road dust contributed almost as
much PM}Q and coarse particulate mass at the Hayden Jail as ore dust, but
over ten times less than the ore dust source at the Garfield site.  The only
other sources contributing more than 1 percent of the suspended particulate
PM^o mass were sources of secondary sulfate and the copper ore tailings
pile, both of which contributed about 5 percent of the particulate mass.

     On days when there were high winds from the east (Regime 1), ore dust
(DUSTO and CUOR) was the dominant source of PM^Q mass at both sites and
responsible for about 70 percent of the mass with about 25 percent of the
mass unexplained.  On low wind speed days when the wind was from the south
(Regime 2), ore dust was still the largest source of PM^Q mass at the
Garfield site, but Kennecott road was the largest source at the Hayden Jail
site.  On days with moderate wind speeds from the southeast (Regime 3), ore
dust was the largest source of PMjg mass at both sites as it was on Regime 1
days, but on Regime 3 days, with moderate wind speeds, Kennecott Road
contributed almost as much mass as the ore dust source category.  On the one
day that represented Regime 4 when the wind was from the west at moderate
wind speeds, the major source impacts at the Hayden Jail was Kennecott road
dust which was responsible for 50 percent of the PM^Q mass, four times more
than the impact from ore dust.  Under this meteorological regime, calcium
rich sources (CAO and GYPSU) were responsible for over 15 percent of the
PM^Q mass at the Hayden Jail site, but only 2 percent of the mass at the
Garfield site.

     Secondary sulfate contributed about 3 or 4 ug/m3 to PM^Q particulate
mass and was relatively independent of meteorological regime suggesting a
regional background source of this species.
                                    3.8

-------
3.2  CMS Model Validation

     Chemical mass balance (CMB) receptor model calculations are performed
on individual filter data sets.  These calculations yield the most probable
source contributions based on ambient and source aerosol chemistry.  Because
these calculations are generally made independently of meteorological
characteristics, plant operating schedules, etc., the validity of the CMB
results can be evaluated by comparing them with these independent airshed
and source characteristics.

     The objective of this section is to evaluate the source apportionment
results relative to the complete airshed data base and model evaluation
statistics. _It is clear from the results that not all of the source
profiles have been characterized nor have the source profile uncertainties
been adequately defined by replicate source sampling over an extended period
of time.  The impact these limitations have on the source contribution
results are discussed in the following subsections.

3.2.1  Quality of CMB Results

     The quality of each CMB source apportionment calculation is evaluated
on the basis of the percent of total mass explained, R-square, chi square,
t-statistic, source uncertainty clusters, elemental ratios, and residuals.
Results of the CMB statistical analysis are presented in the NEA Report.

     The total deposit mass was reasonably well explained considering that
secondary sources of carbon and nitrate were not included.  In addition, a
source rich in copper has clearly been omitted.  The average unexplained
coarse particle mass was only 1 percent at the Hayden Jail site, but was up
to 10 percent at the Garfield site.  The unexplained fine particle mass at
both sites averaged about 20 percent.  However, the average R-square value
was 0.97 and the lowest value was 0.82, indicating generally good fits with
the fitting species used.  The reduced chi square values are relatively high
averaging 3.8 for fine particles at the Hayden Jail and 3.6 at the Garfield
site, and 2.1 and 2.9 respectively for coarse particles*  These high chi
square values are due primarily to the use of source profile uncertainties
that do not reflect the true source profile variability as a function of
time or space and/or the omission of one or more sources.  (Although
replicate samples of some sources were obtained, they were collected only a
few hours apart, were for sources making small contributions, and do not
represent the full range of potential source variability over several
months).

     The source profile uncertainty used was the larger of the standard
deviation of the replicate values, the analytical uncertainty, or 10
percent.  The uncertainties for the major elements in the two largest source
categories (Kennecott Road and ore dust), however, were only 10 percent and
represented essentially only one sample.  Most of the t-statistic values
were greater than 2 and generally 4 or greater.

     The elemental ratios and residuals for the fitting elements were, for
the most part, in the acceptable range, but there were some elements that

                                    3.9

-------
consistently exhibited poor fits.  Most notable was copper, which was
removed as a fitting element at the beginning.  Other elements which were
well above their detection limits but frequently underexplained (low ratios,
high negative residual) included SI, K, Mn, and Fe.

     In general, the best quality fits were obtained with filters with
deposits collected when wind speeds were low to moderate and from the south
or southeast.  This group of thirty-six filters represents almost three
fourths of the filters analyzed.  The average coarse particle mass explained
was 100 +_ 2 percent, and the average fine particle mass explained was 84
percent.  The poorest quality fits were obtained for filters collected
during high wind speeds from the east (Regime 1).
            •
     Although the quality of the CMB modeling calculations could be
improved, primarily through more complete source characterization, the major
source contributions are not expected to be significantly different except
for the source responsible for the excess Cu and other underexplained
species.

3.2.2  Meteorological and Site Dependence of Source Contributions

     The meteorological and site dependence of the calculated source impacts
were consistent with the known source-receptor relationships.  For example,
the average ore dust source contribution to PM^o was about three times
greater at the Garfield site than at the Hayden Jail as expected from the
close proximity of the Garfield site to this source.  On the other hand, the
average contribution of dust from Kennecott Road that runs behind the west
side of the Jail was about four times greater at the Jail than at the
Garfield site.  In addition, when strong winds were out of the east, the
Kennecott Road dust contributions were minimal or below detection limits,
and the ore dust contributions were high at both sites.  The highest ore
dust contributions to the coarse particle fraction at site 2 all occurred
during regimes when the wind was from the east or southeast and the
contributions were more than double the average contributions during the
other two wind regimes.

     The source contributions from Ca rich sources such as lime or gypsum
were also significantly higher at the Hayden Jail as expected from the
location of an old lime/gypsum facility just up the hill from the Jail.

     Copper ore tailings dust impacted the sites only twice and both times
the impacts were at the Garfield site.  One of the days was a day when the
Hayden Jail sampler was not operating.

     The impact of fine particle sources such as secondary sulfate,
transportation, and the smelter fugitive process emissions were quite
similar at both sites.

3.2.3  Particle Size Dependence

     The particle size dependence of the source contributions was consistent
with the source emissions characteristics.  For example, about 90 percent of

                                    3.10

-------
the ore,  road and tailings dust was apportioned in the coarse particle
fraction, and about 95  percent of the secondary sulfate was apportioned to
the fine  particle fraction.  This latter point is significant since the
average fine to coarse S ratio was about 1.2, and much of the coarse
particle  S was associated with ore dust.

     The  average transportation (exhaust) impact was small (about 0.2 ug/m^)
and resolved only in the fine particle fraction.

     The  contributions form the smelter's process fugitive emissions (slag
skimming, matte tap, converter and slag pour) were generally quite small
(about 0.8 ug/m3).  Although only 75 percent of these source impacts were in
the fine  particle size fraction, 90 percent to 95 percent of the source
emissions were in the fine fractions suggesting the possible over estimating
of the coarse particle contribution from these source categories.

     It should also be noted that the ore dust profile was developed from
settled dust collected in the ore crusher building.  Thus, the observed low
F/C ratio for this source may be perturbed by the enrichment of coarse
particles in the settling process.

3.2.4  Source Variability

     During a twelve hour period in December and January, when the smelter's
furnace operations were shut down, process fugitive emissions associated
with the  slag skimming, matte tapping, and converter contributed only 0.31
and 0.42  ug/m^ to PM^o on tne 9th and 7th respectively, about half of the
study average of 0.8.

     The  days with high fine to coarse ratios were characterized by lower
than normal contributions from coarse particle sources rather than increased
impacts from fine particle sources which were relatively small contributors
and reasonably constant throughout the study period.

     The  CMB model results on the three days that overlapped the days when
source samples were collected were among the highest quality*  The average
R-square  value was 0.97 +_0.01, and the average chi square value was 2.1 +_
0.7.  The average fine particle chi square value was 1.7, while the coarse
particle  chi square value was 2.5.  The percent coarse particle mass
explained was 100 percent, and the fine particle mass explained was 83
percent.

     Poor fitting of Cu in many airsheds is not considered significant
because of the potential contribution from the sampler pump motors.  The
problem is probably more general in urban airshed, however, because of the
large number of potential electrical sources of Cu.  In this particular
case, the excess Cu is not thought to be due to either of the above sources
because of the low population density, good ventilation at the sampling
sites, and absence of high volume samplers at the Garfield site.

     Source profiles for the fugitive process emissions are based on samples
collected on these three days.  The significance of this close overlap of

                                    3.11

-------
ambient and source sampling is most clearly reflected in the excellent
quality of the fine particle fits and ratios close to 1.0 for the elements
most abundant in the fine particle fugitive dust emissions such as Zn, As,
and Pb.

     The poorer quality of fit obtained for the coarse particle fraction on
these three days is quite likely due to the contribution of other dust
sources either not sampled or included in a composite, or variability in the
ore dust emissions from the crusher and/or the conveyor belt.  Although a
variety of smelter yard dust and enriched ore samples were collected during
the last week in January, many of the samples were combined into one
composite.  In addition, even though samples of dust in the ore crusher
building were collected, filter samples collected in the building were
invalidated because of material loss in shipment.  The bulk dust collected
from within the building and later aerosolized in the laboratory represented
the accumulation of dust over an extended period of time and was not
necessarily representative of the emissions on these three days.

3.2.5  Major Source Impacts

     -  Ore Dust

        Ore dust was the largest contributor to fine, coarse and PMjg
particles at both sites.  The fine to coarse source contribution ratios were
typical of the low fine to coarse particle ratio observed with the source.
The impacts calculated for this source are consistent with the spatial
relationship of the two receptors relative to the source and the impacts
during the five specified meteorological regimes.  The largest source of
uncertainty in this source contribution estimate is due to the potential
variability in the profile for this source and the lack of replicate samples
to establish this variability.

     -  Road Dust

        The second largest contributor of PM}Q and coarse particles was road
dust.  This source impact, like the ore dust source category, was consistent
with particle size, source receptor, and meteorological regime
stratification relationships.  Similarly, the largest source of error in
this source contribution estimate is due to the limited definition of the
variability of the emission along this road, as well as the potential
influence of a missing or inadequately characterized source.

     -  Secondary Sulfate

        The secondary sulfate source contribution estimate is based on the
assumption that any S not explained by primary sources is secondary sulfate.
Even though the fine and coarse particle S ration was close to 1.0, the
apportioned secondary sulfate was highly concentrated in the fine particle
fraction as would be expected by the nature of this source.  In addition,
the impact from this source was relatively constant and independent of
meteorological regime and receptor location which suggests a source related
to the background air mass.

                                    3.12

-------
     -  Ore Tailings

        The copper ore tailing impacts were observed only on the highest
wind speed days when gusts approached 24 mph which is consistent with the
origin of this dust.  This source impact was observed above its detection
limit only in the coarse particle fraction and not at the Hayden Jail site.

     -  Lime and Gypsum

        The lime and gypsum source contribution estimates were also largely
coarse particle impacts.  Lime and/or gypsum either is currently being used
at the smelter or was in the past.  An old lime or gypsum plant or storage
area can be seen from the Hayden Jail.  The impact was largest at the Jail.
It is also likely that some of the material from the old facility on the
hill above the Jail has washed down the gully passing behind the Jail
enriching the general area with calcium.  A possible but unlikely source of
gypsum is a truck dump and rail car loading area just south of Winkleman.

     -  Process Fugitive Emissions

        This category includes process fugitive emissions form slag
skimming, matte tapping, converter operations, and slag pouring.   The
relative impacts from these sources may vary substantially over a twenty-
four hour period, but the individual source impacts are easily resolvable
when the impacts are substantially above their detection limits.  In this
particular study, the impacts were reasonably small and represent only 1 -
2 percent of the PM^Q mass.  This source category, however, is responsible
for 80 to 100 percent of the Pb mass and the toxic species of As and Cd.
Although most of the impact from this source category was attributed to the
fine particle fraction, the fine to coarse particle source impact ratios
were higher than expected form the source profile characteristics.  This is
thought to be due in part to the poor source signal to noise ratio in the
coarse particle size fraction and the potential omission or mis-
characterization of a coarse particle source.  As a result, it is likely
that the coarse particle source contribution estimates for this source
category are overstated.

     -  Smelter Yard Dust

        This source category, which is also called composite plant road dust
(PLNTRD or COMPR), is a mixture of five dust samples collected within the
smelter.  This composite sample has the highest concentration of Cu and it
is assumed that some of the individual dust samples included in this
composite source category have even higher Cu concentrations.  It is
considered likely that one or more of the dust sources included in this
composite are responsible for more of the fugitive yard dust coming from the
smelter than others included in the composite.   These possible dust sources
would have higher Cu concentrations than reported for the composite and are
possibly enriched in species such as Fe and Mn.  This assumption is
consistent with the impacts and chemistry on meteorologically regime
stratified days.

                                    3.13

-------
        Sources Not Included

     Several sources of emissions  from the smelter were not characterized or
included in the CMB model calculations.  These sources included the main
stack, acid plant stack, Cu anode  blowing, bell caps, and Cu pouring.  The
impacts on local receptors from all of these sources were considered to be
small or at least less than the process fugitive emissions measured.
Consistent with this assumption is the relatively low impact from the sum of
the main process fugitive emissions sources sampled and the reasonably good
fits obtained for most of the fine particle filters.  In addition, the low
fine to coarse particle ration for such elements as Cu suggests that the Cu
anode blowing and pouring, which is expected to emit primarily fine Cu
particles, is relatively unimportant.

3.2.6  Large Chi Square Values and Unexplained Mass

     It was not possible to obtain reduced chi square values close to 2.0
for several of the data sets.  The worst case was for the fine particle
sample collected on November 22, 1986, at the Hayden Jail where the lowest
reduced chi square value that could be obtained was 10.7.  The element
primarily responsible for the high chi square value is K.  The source
profile library does not show a source rich in K.  The only way the chi
square value can be reduced is to eliminate K from the fitting.   This may
be a case where one of the smelter emissions mentioned above that were not
characterized is making a substantial contribution to the fine particle
fraction even though the average wind direction was from the west.  Silicon,
in addition to K, were also poorly fit in the fine fraction on this day at
the Garfield site, and a similar high chi square value obtained (6.2).  In
contrast, the chi square values for the corresponding coarse fraction were
2.4 at the Jail, 1.1 at Garfield.  in addition, the elemental ratios for K
and Si were close to 1.0 in the coarse fraction.

     In this particular example, it is likely that either an uncharacterized
smelter source emission impacted the fine particle filter, a source such as
the mine west of the smelter impacted the site, or one of the fine particle
source profiles was substantially altered.  This was the only day analyzed
in which the average wind direction was from the west (261°).

     It is likely that similar variations in the source profiles and/or
source impacts are responsible for the other samples for which high reduced
chi square values were obtained.

     -  Unexplained Mass and Elemental Concentrations.

        The unexplained mass was largest for the fine particle fraction
which is typical of most airsheds.  In this study, most of the unexplained
fine particle mass is thought to be associated with secondary organic carbon
and nitrate, as well as water and sources rich in carbon but not resolved.
In some cases, some of the unexplained mass is probably associated with the
impact from uncharacterized sources.  In the coarse particle size fraction,


                                    3.14

-------
the unexplained mass is likely due to the omission of a smelter  yard  dust
source.
                                    3.15

-------
4.0  RECONCILIATION OF RECEPTOR/DISPERSION MODEL RESULTS

     The development of control strategies needed to attain/maintain the
PM^o standards are dependent on the ability of available modeling techniques
to properly predict the contributions from each source at selected model
receptors.  The PMm SIP Development Guideline recommends that both Chemical
Mass Balance (CMS) and Dispersion Models (DM) be used to complete the source
apportionment.  Because these techniques use different inputs and
calculation techniques, the resulting predicted PM^g concentrations cannot
be expected to show perfect agreement.  The differences may be negligible,
but on occasions they may be large enough to raise questions concerning
which model is correct.  A systematic method for examining and reconciling
the model estimates is contained in the EPA's Protocol for Reconciling
Differences Among Receptor and Dispersion Models.  The steps outlined in the
protocol were used to reconcile the differences between the CMB and DM
analyses performed for Hayden.

4.1  Preliminary Dispersion Modeling

     Using the DM data base established for the study area (see Section
2.3), preliminary DM estimates were made.  These estimates were then
compared with the CMB results in order to determine whether all major source
groups were adequately defined in the emission inventory.

     For Hayden, preliminary dispersion modeling was performed for the 15
days which were evaluated using the CMB.  Site-specific meteorology for the
days selected for analysis was used to predict 24-hour average PM^g impacts
from the sources contained in the emission inventory.  The dispersion
modeling revealed discrepancies between the results from the CMB and DM
analyses which needed to be reconciled before modeling of potential control
strategies could be performed.

4.2  Model Reconciliation

     The Protocol for Reconciling Differences Among Receptor and Dispersion
Models (henceforth referred to as the Protocol) provides an eight step
approach which involves the comparison of model results, reverification of
model input data and the refinement of inputs to both models.  The general
protocol for resolving the differences between the two modeling approaches
is summarized in Figure 4.1.  This protocol provides the basic framework for
the reconciliation of the PM^g modeling analyses for Hayden.

     The PM^o emissions in the Hayden area were initially grouped into
source categories which generally corresponded with those used in the CMB
analysis.  Dispersion modeling impacts from these source groups provided for
a quick check against the CMB predictions.  The source groups used in the
initial ISCST modeling are given below.  The locations of the point and area
sources, along with the monitor locations, can be found in Figure 2.1 and
in Section 2 of this report.

          - Automobile Exhaust
          - Roadway Dust

                                    4.1

-------
                                                              FIGURE 4.1
FLOW DIAGRAM FOR MODEL RECONCILIATION
             CMB RESULTS
   CORRECT MODEL
  INPUTS AND RERUN
                NO
 MODEL
 INPUTS
CORRECT
   T
         OM RESULTS
                                           YES
                       REFINE CMB MODEL
                       INPUTS AND RERUN
                           MODEL
                           RESULTS
                           AGREE
                 YES
                       REFINE DM MODEL
                       INPUTS AND RERUN
                                            YES
                               NO
                      FURTHER REFINEMENT
                       TO CMB AND/OR OM
                           INPUTS
                               NO
                           MODEL
                          RESULTS
                           AGREE
                             7
                 YES
                               NO
                     DETERMINE WHICH MODEL
                     IS APPROPRIATE FOR USE
                      IN CONTROL STRATEGY
                        DEVELOPMENT
                   USE DM FOR CONTROL
                  3TATEGY DEVELOPMENT

-------
          - Copper Ore Tailings
          - Smelter Building Fugitives
          - Slag Dump
          - Copper Ore
          - Ore Crusher
          - Lime Handling
          - Gypsum
          - Secondary S04
          - Other

The "other" category contained sources of PM^Q from woodburning,
locomotives, windblown dust and the main ASARCO stack.  The "secondary 804"
category was.a single point source located at the ASARCO smelter building.
This source was not used so much for predicting sulfate impacts as for
determining if there was any wind direction correlation with measured
secondary sulfate levels.  If a strong correlation between easterly (from
the east) winds and high sulfate levels was revealed, it could be reasonably
assumed that the sulfate emissions were emanating from the most likely
source, the smelter building.  Lack of such a correlation would suggest that
more complex transport/chemical processes led to the measured sulfate
levels.

     Because an annual-average emissions inventory was used in the DM,
comparison of model results on a day-by-day basis was not appropriate.  The
results were aggregated into groups of days characterized by similar
dispersion meteorology as recommended in the Protocol.  The CMB results were
aggregated into five groups, three of which contained more than one day.
The three multi-day groups, along with a group containing all the days
examined, were used to reconcile the differences between the CMB and DM
estimates.

     The large discrepancies between the initial dispersion modeling and CMB
results indicated that the protocol outlined in Figure 4.1 should be used to
reconcile the differences.

     Examination of the input data to the ISCST model revealed that several
of the point sources and the Garfield monitor were not properly located
within the modeling domain.  'These errors were corrected and the dispersion
modeling was performed once again using the revised source/receptor
coordinates.

     The revised dispersion modeling results were again compared to the CMB
results.  The changes in source/receptor locations improved model agreement
on some days while causing greater disagreement on others.  Overall, no
significant improvement was detected.  The ISC model was overpredicting
concentrations, most significantly at the Garfield monitor.  The tendency of
ISC to overpredict is illustrated in Table 4.1.

     The results presented in Table 4.1 illustrate the major discrepancies
between the results of the two modeling approaches.  The larger ISC
overprediction at the Garfield monitor is mainly attributable to the
contributions form the ore crusher and copper ore groups.  The ISC results

                                    4.3

-------
                                 Table 4.1
                        Preliminary Modeling  Results
                Hayden PM^Q Modeling - Model Reconciliation
Date: All Days
Source Group
All Sources
Auto Exhaust
Roadway Dust
Tailings
Smelter Building Fug.
Slag Dump
Copper Ore
Ore Crusher
Lime Handling
Gypsum
Secondary 864
Other*
Total Measured
24-hr Impact (ug/m^)
Garfield
CMB
104.51
0.23
4.69
5.29
0.86
0.06
0
73.75
0.34
1.63
4.11
13.44
104.51

ISC
398.96
0.42
13.46
2.35
1.02
2.17
19.65
332.63
1.06
1.88
5.04
16.92

Jail
CMB
59.49
0.19
20.71
0
1.04
0.03
0
27.72
0.57
2.94
4.03
2.29
59.49

ISC
85.77
0.39
23.21
1.51
1.11
2.87
10.34
31.51
1.65
2.39
5.50
5.29

* ISC: Woodburning, Locomotives, Windblown Dust, Stack
  CMB: Unexplained
                                    4.4

-------
also indicate that significant PM^g contributions at both monitoring sites
arise from roadway dust and sources not identified by receptor modeling
techniques.  The predictions from the copper ore tailings and the ASARCO
slag dump also reveal disagreement.  The DM results over predict the CMB
results further if background PM^Q levels (not shown) are included.  Table
4.1 also shows that, while the magnitudes do not show good agreement, both
models identify roadway dust and crusher-related activities as the major
source contributors to the PM^g levels at the monitoring sites.
     Certain details of the model comparison which indicated adjustments to
model inputs should be made are not shown by the 15 day composite numbers in
Table 4.1.  For example, the DM did not show conclusively that secondary
sulfate impacts were associated from a single source at the ASARCO smelter
building.  As a result, the 804 source was eliminated from further
evaluation and secondary SO^ levels were assumed to be contained in the
background PM^g concentration.  The CMB results showed that impacts from the
copper ore tailings were seen on only two days at the Gar field monitor.
Review of the 15 days of meteorological data revealed wind speeds which were
generally below the threshold needed to generate emissions from the tailings
pond.  This led to the elimination of the copper ore tailings sources from
all days except the two days where tailings impacts were estimated by the
CMB.  Further examination and reconciliation of the impacts from the roadway
dust, slag dump, copper ore, ore crusher and locomotive exhaust source
categories was deemed necessary based on the preliminary modeling.  A
discussion of the refinement of the estimates from each of these categories
is given below.  Furthermore, an annual-average PM^g background
concentration of 16 ug/m^ measured at Organ Pipe National Monument in
southcentral Arizona was determined for incorporation into subsequent
dispersion modeling.

     The Protocol recommends that the reconciliation process begin with
refinements to _ the CMB analysis.  The CMB results showed a generally good
characterization of the major PM^g contributors in Hayden (see Section 3).
Conversely, the emissions estimates used in the initial dispersion modeling
contained a high degree of uncertainty which are embodied in the associated
DM concentration" estimates .  Therefore , it was determined that
reconciliation of the model estimates would be accomplished through the
refinement of DM emission inputs and that no refinement of the CMB results
was necessary.  The source characterization and CMB analyses were discussed
with NEA, Inc. throughout the model reconciliation process.  Clarification
of the quality of source samples used to characterize a source or source
group as well as CMB predictions provided the basis for the refinement of
the emission estimates used in the dispersion modeling.

     The Protocol states that acceptable model agreement is achieved when
the CMB and DM results agree within one standard error.  Because the results
from both modeling techniques have an associated uncertainty, acceptable
model agreement would occur when the ranges of the predicted concentrations
(defined by the calculated impact and the standard error) overlap.  For
example , a CMB-predicted concentration of 100 +_ 5 ug/m3 and a DM-predicted
concentration of 120 +_ 36 ug/ra^ would show acceptable agreement because the
CMB concentration range of 95 ug/m^ to 105 ug/nH overlaps the DM range of 84

                                    4.5

-------
ug/m3 to 156 ug/m3.  The Protocol recommends assigning a standard error of
30 percent to estimates made using EPA guideline dispersion models.

     The refinements made to reconcile the differences in the modeled
results are presented here by source category.

     Ore Crusher/Copper Ore.  The primary contributor to PM^Q
     concentrations identified in the preliminary results of both models was
     the ore crusher.  The influence of this source is more pronounced at
     the Garfield monitoring site than at the Hayden Jail site, due to the
     close proximity of the crusher to the Garfield monitor.

     The wide range of DM over- and underprediction at both sites indicated
     that:

          -  source characteristics (i.e., stack parameters) were incorrect;
          -  single point source characterization was incorrect;
          -  emission rate was incorrect; or
          -  a combination of all of the above.

     The stack parameters used in the modeling were reviewed and it was
     found that the exit velocity from the crusher stack was significantly
     underestimated.  Photographs taken during a site visit also showed
     emissions from a transfer point between conveyors which is  located to
     the east of the crusher building.  This area was identified as a source
     of PM|Q and a portion of the crusher emissions were assumed to be
     emitted from this point.  The ISC model was exercised once again using
     the revised exit velocity at the crusher stack and emissions from the
     conveyor transfer point.  The resulting concentrations showed much
     better agreement -with the CMB results, although the model continues to
     overpredict it was apparent that the estimated emissions from the
     crusher were too large.  Photographs of the crusher building showed
     that the crusher building was more enclosed than originally thought.
     Therefore, the amount of fugitive emissions contained in the original
     estimate were reduced.

     The results also showed that for many cases, combined impacts from the
     crusher and copper ore sources exhibited generally good agreement with
     the CMB ore crusher impacts.  Furthermore, the CMB results did not show
     copper ore impacts on any of the 15 days evaluated.  Review of the
     source sampling data showed that the grab sample used to create the
     copper ore source profile was not of high quality.  The lack of any
     CMB-predicted copper ore contributions was most likely due to the use
     of a source fingerprint which was not representative of the source(s).

     The copper ore impacts in the initial ISC modeling appeared to be
     excessively high, with maximum concentrations in excess of 90 ug/m^.
     Because of the large uncertainty in estimating the emissions from the
     train ore dump and conveyor unloading activities at the ore storage
     area, the emissions from these sources were reduced.

     Because the characteristics of ore storage and ore unloading sources

                                    4.6

-------
(copper ore group) are probably very similar to those from the crusher
and conveyor, the crusher and copper ore groups were combined for
further reconciliation.

The model was run once again with the revised emission rates and the
revised "copper ore" source group containing the copper ore sources and
the crusher-related sources.  The results showed generally good
agreement with the CMB results.  For example, the highest predicted CMB
concentration was 179.33 ug/m3 while ISC predicted a maximum
concentration of 273.78 ug/m3.  The 15-day average impacts also agreed
favorably with an average concentration of 73.75 ug/m3 predicted by CMB
and a value of 61.47 ug/m3 estimated by ISC at the Garfield site.  For
this case, the ISC results were within one standard error of the CMB
results.  The models did not agree quite as well at the Jail monitor
where the 15-day ISC-predicted impact of 12.10 ug/m3 compared with a
CMB-predicted concentration of 27.72 ug/m3.

Roadway Dust.  The predicted roadway dust concentrations showed
generally good agreement, especially at the Jail monitor.  However, on
specific days where easterly winds were predominant, CMB results
indicated no roadway dust impacts at either one or both sites while the
DM predicted impacts at both monitor locations.  The reason for the
discrepancies was determined to be associated with the allocation of
roadway emissions to area sources for dispersion modeling purposes.
The emissions were reallocated to smaller area sources and/or point
sources within the area source grids in close proximity to the monitor
locations.  These changes did not significantly improve model
agreement.  It was therefore determined that the initial area source
allocation provided a generally good representation of roadway dust
emissions in Hayden.  For example, the maximum predicted roadway dust
contributions were 46.82 ug/m3 (CMB) and 39.90 ug/m3 (ISC).  The 15 day
composite values also showed good agreement with average impacts of
23.21 ug/m3 (ISC) and 20.71 ug/m3 (CMB) predicted at the Jail site.
These values agree within the one standard error prescribed in the
Protocol.  Model agreement was not as good at the Garfield site where
the 15 day composite value of 13.46 ug/m3 from the ISC modeling
compared with a CMB-predicted concentration of 4.69 ug/m3.

Copper Ore Tailings.  The wind speeds during the 15 days of ambient
sampling indicated that negligible impacts from the tailings piles
would be expected.  This was corroborated by the CMB results where no
tailings impacts were predicted on 13 of the 15 days.  However, two
days show impacts in excess of 30 ug/m3 at the Garfield monitor.

Because emissions from the tailings pile are a function of wind speed,
the meteorological data for these two days were examined in an attempt
to explain the high impacts.  Wind measurements in the vicinity of the
tailings pond indicated speeds above the nominal wind speed emission
threshold for emissions of 11 miles per hour.  The threshold speed for
tailings emission generation was determined from on-site experiments
using a portable wind tunnel.  Thus, both days were evaluated with the
ISC model using variable hourly emission rates which were based on the

                               4.7

-------
     measured wind speeds.  The results showed negligible impacts on both
     days.  Further examination of the meteorological data revealed
     generally southeast flow on these days, that transported the majority
     of the emissions away from Hayden to the northwest.  The discrepancies
     between the models could be attributable to the inability of the
     Gaussian-based dispersion model to address short-term phenomena such as
     wind gusts or dust devils.  Such micro-meteorological events could
     transport tailings to the Gar field monitor, but not be adequately
     addressed by the model which uses a single wind speed and direction to
     represent an entire hour.

     Locomotive Exhaust.  Examination of the preliminary modeling results
     indicated that the initial locomotive exhaust emissions were
     overestimated.  While the CMB analysis did not examine locomotive
     impacts specifically, it was assumed that any contributions would be
     contained within the "unexplained" category.  On specific days, initial
     DM impacts from the locomotive sources greatly exceeded the CMB
     "unexplained" impacts.  The emissions calculations were examined to
     reveal that the train activity levels used to determine the locomotive
     emissions were overestimated by a factor of four.  Therefore, the
     emissions were reduced by 75 -percent and the model was rerun.

     Slag Dump.  Initial DM predictions generally exceeded the CMB
     estimates, sometimes very significantly.  The slag dump emissions were
     originally calculated using a roughly-estimated slag dumping rate of
     20 ton/hour.  This estimate was refined to approximately 10 ton/hour.
     Therefore, the original slag dump emissions were reduced by 50 percent.

4.3  Summary of Model Reconciliation Analysis

     The results of the receptor/dispersion model reconciliation analysis
were summarized in order to determine the modeling technique(s) to be used
for the control strategy development portion of the SIP.

     Based on the refinements to the DM data base for the Hayden area
described above, the models were recompared and found to show generally good
agreement.  Table 4.2 presents the results from the 15 day composite
groupings while Tables 4.3, 4.4 and 4.5 present comparisons for days with
similar dispersion modeling.
     Overall, the ISC model  tended to predict total PM^Q concentrations
which were in good agreement with the CMB estimates.  Table 4.2 shows that
copper ore impacts exhibit good agreement at the Garfield monitor while
roadway dust concentrations  agree very well at the Jail site.  The ISC model
tended to underpredict roadway dust impacts at the Garfield site and the
copper ore impacts at the Jail monitor.  The impacts from the other source
groups showed generally good agreement at both monitor locations.

     Tables 4.3, 4.4 and 4.5 illustrate how the models compare for days
exhibiting similar meteorological conditions.  Table 4.3 shows that with
predominantly easterly winds, ISC overpredicts roadway dust impacts and
overpredicts copper ore concentrations at the Garfield site.  The southerly

                                    4.8

-------
                                     Table 4.2
                              Final Modeling Results
                    Hayden PMjQ Modeling - Model Reconciliation
                                     All Days
24-hr Impact (us/m^)
Garfield
Source Group
Auto Exhaust
Roadway Dust
Tailings
Smelter Building
Slag Dump
Copper Ore
Lime Handling
Gypsum
Secondary 504
Other*
Background
All Sources
Total Measured
CMB
0.
4.
5.
Fug. 0.
0.
73.
0.
1.
4.
13.
^^•••M
104.

23
69
29
86
06
75
34
63
11
44

57

+ 0.12
+ 2.46
+ 3.60
+ 0.66
+ 0.06
+ 14.73
-l- 0.30
+ 0.48
+ 0.43
+. 5.03


104.51
ISC
0.42
13.46
0.01
1.02
1.08
61.47
1.06
1.88
—
4.82
16.00
101.22

+ 0.
+ 4.
+ 0.
+ 0.
+ 0.
+ 18.
+ 0.
1 0.

1 1.



13
04
00
31
32
44
32
56

45



Jail
CMB
0.19
20.71
0
1.04
0.03
27.72
0.57
2.94
4.03
2.29
—
59.49

-I- 0.14
+ 4.69
+ 0.00
+ 0.66
+ 0.03
+ 7.51
+ 0.57
+ 0.71
+ 0.58
± 1.66


59.49
ISC
0.39
23.21
0
1.11
1.43
12.10
1.65
2.39
—
2.30
16.00
60.58

+_
t
t
±_
+_
t
t
+_

+



0.12
6.96
0.00
0.33
0.43
3.63
0.50
0.72

0.69



* ISC:  Woodburning,  Locomotives,  Windblown Dust,  Stack
  CMB:  Unexplained
                                        4.9

-------
                                 Table 4.3
                           Final Modeling  Results
                Hayden PM^g Modeling - Model Reconciliation
                                  Regime 1
Source Group
Total Measured
                                    24-hr Impact
                                          Garfield
                                  CMB
                                                        ISC
Auto Exhaust
Roadway Dust
Tailings
Smelter Building Fug.
Slag Dump
Copper Ore
Lime Handling
Gypsum
Secondary SO^
Other*
Background
All Sources
0
1.03
22.94
0.99
0.03
87.27
1.49
0.13
3.17
40.81
—
157.85
+ 0.00
+ 1.03
+ 11.57
+ 0.99
+ 0.03
+ 47.28
+ 1.19
+ 0.13
+ 0.93
+ 5.11


0.38
11.54
0.06
1.31
0.05
120.69
0.58
1.49
—
4.32
16.00
156.42
+ Ooll
+ 3.46
+ 0.02
+ 0.39
+ 0.02
+ 36.21
+ 0.17
+ 0.45

t 1.30


                                            157.85
* ISC: Woodburning, Locomotives, Windblown Dust, Stack
  CMB: Unexplained

Regime 1 = 12/29/86, 1/11/87, 1/12/87
Wind Direction: East
                                    4.10

-------
                                     Table 4.4
Hayden
                              Final Modeling Results
                                Modeling - Model Reconciliation
                                     Regime 2
Source Group
                                      24-hr Impact
                                 Garfield
                                            Jail
      CMB
ISC
CMB
ISC
Auto Exhaust
Roadway Dust
Tailings
Smelter Building
Slag Dump
Copper Ore
Lime Handling
Gypsum
Secondary S04
Other*
Background
All Sources
0
3

Fug. 1
0
42
0
2
4
3

58
.49 +
.39 +_

.19 +
.02 +
.75 +_
+
.29 +
.81 +
.82 +_
••••^^B
.77
0.30
2.57

1.10
0.02
5.44
0.00
0.77
0.73
2.75


0
12
•»
0
0
27
0
1

4
16
63
.40 +
.66 +_

+_
+
.94 +
.21 +
.33 +_
—
.59 +
.00
.13
0.12
3.80

0.00
0.00
8.38
0.06
0.40

1.38


0.29
18.50
—
1.10
0
11.01
0
2.41
4.77
3.10
—
41.19
+ 0.29
+_8.00

+ 0.97
+ 0.00
+ 1.39
+ 0.00
+ 1.36
+ 0.65
+ 1.38


0.44
23.07
—
0.01
0.01
11.84
0.21
3.07
—
2.50
16.00
57.15
+_
+_

+_
+_
+_
+_
+_

+_


0.13
6.92

0.00
0.00
3.55
0.06
0.92

0.75


Total Measured
         58.77
                  41.19
* ISC: Woodburning, Locomotives, Windblown Dust, Stack
  CMB: Unexplained

Regime 2 = 12/9/86, 12/18/86, 12/24/86, 1/7/87, 1/30/87
Wind Direction: South
                                        4.11

-------
winds characterizing Regime 2  (Table 4.4) show excellent agreement between
the CMB and ISC (excluding background) at the Jail monitor and good
agreement at Garfield.  Table  4.5 shows that southeasterly winds result in
excellent agreement at both monitors for roadway dust impacts and relatively
poor agreement for copper ore  impacts.
     Linear regression analyses of total PM^g impacts from CMB and ISC were
performed for both the preliminary and final modeling.  Figure 4.2 presents
the total predicted PM^Q from the CMB and final ISC modeling analysis and
the results of the regression analysis.  The preliminary modeling results
(not shown) showed poor agreement with correlation coefficients of -0.09
(Garfield) and -0.23 (Jail).  Revisions to the ISC inputs described above
resulted in a dramatic improvement in model agreement at the Garfield
monitor location.  The correlation at the Jail site was not, however,
significantly improved.

     Based on the analysis described above, the differences between the
results of the two models which were indicated by the initial modeling were
satisfactorily reconciled to justify the use of the ISCST dispersion model
in the development of PM^o control strategies at Hayden.
                                    4.12

-------
                                     Table 4.5
Hayden
                               Final  Modeling  Results
                                Modeling - Model Reconciliation
                                      Regime 3
Source Group
                                      24-hr Impact (ug/m3)
                                Garfield
                                             Jail
      CMB
ISC
CMB
ISC
Auto Exhaust
Roadway Dust
Tailings
Smelter Building
Slag Dump
Copper Ore
Lime Handling
Gypsum
Secondary 304
Other*
Background
All Sources
0
10

Fug. 0
0
98
0
1
3
1

iT7
.25
.46

.70
.18
.18

.85
.90
.54

704
+_
+_

+_
+_
+_
+_
+_
+_
+_


0.25
6.14

0.63
0.18
27.74
0.00
1.10
0.97
3.64


0.50
10.77
—
1.23
0.70
34.97
1.03
2.67
—
4.46
16.00
72.33
+ 0.15
+_ 3.23

+ 0.37
+ 0.21
+ 10.49
+ 0.31
+_ 0.80

+_ 1.34


0
23.98
—
1.28
0.01
49.61
0
3.26
3.92
1.48
—
80.55
+ 0
t 5

+ 1
+ 0
+ 12
+ 0
+ 1
+ 1
-i- 2


.00
.97

.01
.01
.51
.00
.00
.01
.14


0.34
22.28
—
2.53
3.29
14.62
2.93
1.56
—
1.95
16.00
65.50
+ 0.10
+.6.68

+ 0.76
+ 0.99
+ 4.39
+ 0.88
+_0.47

+_0.59


Total Measured
        117.04
                   80.55
* ISC: Woodburning, Locomotives, Windblown Dust, Stack
  CMB: Unexplained

Regime 3 = 12/1/86, 12/4/86, 1/4/87, 1/27/87, 1/30/87
Wind Direction: Southeast
                                        4.13

-------
                                                       FIGURE 4.2
            COMPARISON OF CMB AND FINAL ISC

      MODEL PREDICTIONS OF PM   CONCENTRATIONS
   300 p
6


3  200

CO
t-
o

a.
2

O
Ul
Q
UJ
oc
a.  100

o
co
GARFIELD
                               JAIL
r-0.61
                  SLOPE  0.91  0.15

               INTERCEPT 18.03 52.02

                       r«0.61  0.24
                                    r.0.24
                      100
                 CMB-PREDICTED IMPACTS
                  200
                300

-------
5.0  BASE YEAR DISPERSION MODELING

     Prior to the development of the emissions control strategies, maximum
24-hour and annual-average PM^o concentrations associated with the base year
emissions need to be determined.  This is generally done using dispersion
modeling of the base year (1986) PM^Q emission inventory.  The Guideline on
Air Quality Models (Revised) provides guidance on dispersion modeling of
particulate matter and lists preferred models for this purpose.

     For Hayden, the ISCST model was selected to calculate the 24-hour and
annual-average concentrations associated with the "final" 1986 PMjo emission
inventory.  The final inventory used was developed in the receptor/
dispersion model reconciliation analysis presented in Section 4 of this
report.  One year of site-specific sequential hourly meteorological data
collected in 1983 were used as inputs to the model.  A description of the
ISCST model and the meteorological data are presented in Section 2.3 of this
report.  A receptor grid defines the non-attainment area and encompassing
the towns of Hayden and Winkelman was deployed to locate the maximum 24-hour
and annual-average impacts.  The receptors were spaced at 400 m intervals
within a 2 mile x 2 mile area.  Receptors were also placed at the location
of the Jail and Garfield monitor locations which were used in the model
reconciliation analysis.
     The PM^o SIP Guideline Document states that design concentrations can
be based on ambient measurements of PM^o or niodel estimates of ambient
concentrations.  Ideally, modeling estimates using five years of National
Weather Service meteorological or at least one year of on-site data and
three years of air quality measurements should be considered in determining
24-hour design concentrations.  Approximately two years of monitored PM^Q
data and modeling results using one year of site-specific meteorological
data were used to determine the 24-hour and annual design concentrations at
Hayden.
     The PMiQ SIP Guideline Document provides a table look-up procedure for
determining which measured PM^o concentration should be considered in the
determination of the 24-hour design concentration.  This procedure is based
on the total number of measured 24-hour PM^o concentrations and is
sunmarized in Table 5.1.  For Hayden, the highest concentration measured at
the Jail monitor was selected since 57 samples were collected during 1985
and 1986.

     The Modeling Guideline indicates that when 5 years of weather service
data or at least 1 year of site-specific data are used to estimate modeled
impacts, the highest second-highest short-term concentration should be
considered in the design concentration determination process.  If it is
determined that the on-site data may not be temporally representative, the
highest concentration should be used as the design concentration.  Because
one year of site-specific meteorological data were used in the modeling for
Hayden, the highest second-highest modeled 24-hour concentration was
selected for evaluation as the design concentration.
                                    5.1

-------
                                 Table 5.1

             Tabular Estimation of  PM^Q  Design Concentrations
Number of Daily
     Values
Rank of Upper
    Bound
Rank of Lower   Data Point Used for
    Bound       Design Concentration
     <_347

 348 - 695

 696 - 1042

1043 - 1096
     1

     2

     3
      1

      2

      3

      4
Highest Value

Second Highest Value

Third Highest Value

Fourth Highest Value
                                    5.2

-------
     The design concentrations selected for control strategy development are
the highest of the monitored or modeled values.

5. 1  24-hour Design Concentration
     The maximum 24-hour PM^Q concentration measured in Hayden between 1985
and 1986 was 243 ug/m3 while the maximum modeled concentration was 618
ug/m3.  The maximum modeled concentration occurred at the Garfield monitor
site with major contributions from the copper ore sources, roadway dust and
locomotive exhaust emissions.  Based on the selection criteria set forth in
the Guideline Document and the Modeling Guideline, the design concentration
was determined to be 618 ug/m3.  This concentration provides the basis for
the control strategies needed to meet and maintain the 24-hour NAAQS of 150
ug/m3.

5.2  Annual Arithmetic Mean Design Concentration
     The maximum annual arithmetic mean PM^Q concentration measured in
Hayden in 1985 and 1986 was 80 ug/m3.  The maximum modeled concentration was
103 ug/m3.  As with the maximum modeled 24-hour concentration, the maximum
annual-average concentration occurred at the Garfield monitor site with
major contributions from the same sources discussed above.  Thus, the value
of 103 ug/m3 was selected as the annual-average design concentration and
provides the basis for the control strategies needed to meet and maintain
the annual arithmetic mean NAAQS of 50 ug/m3.
                                    5.3

-------
6.0  CONTROL STRATEGY ALTERNATIVES - PM10

     The preceding sections describe the methods and procedures for data
base development, establishment of source-receptor relationships through
receptor modeling, dispersion modeling techniques and determination of base
year PM^Q annual arithmetic mean and 24-hour concentrations.  This section
will discuss the ambient PM^Q concentration reductions needed to achieve the
NAAQS and the emission reduction options available for each source category.


6.1  Emission Reductions Needed to Attain NAAQS
     Section. 6.4 of the PM^Q Guideline Document describes the following
procedure for calculating the total reduction (TR) in pollutant
concentration and the associated emission reductions needed to achieve the
     NAAQS and acceptable toxic levels:

     TR (ug/m3) = PMio Design Concentration - PM^Q NAAQS
If a design concentration is greater than the NAAQS (TR is positive), as is
the case for the Hayden area, a reduction in PMjg emissions is required.
This might be accomplished by reducing the contribution of a single source
or it may require reduction in several individual sources or source
categories so that

     TR (ug/m3) = IISRi (ug/m3)
     where ISR^ is the Individual Source Reduction (as expressed in terms of
     ambient concentration reduction for each source) desired from a source
     or source category i.

These ISRs are generally selected based on many considerations, including
the technical and economic feasibility of achieving a given emission
reduction or additional reduction at that source.  The percent reduction in
emissions (%RE) for a source or source category is given by

     %REi = ISRi (ug/m3)
            ACi (ug/m-i)
     where AC is the ambient"concentration due to the individual source (i)
     or so'urce category as determined through modeling.

     The Hayden base year modeling results indicate that attaining the 24-
hour NAAQS will require a more restrictive control strategy than attainment
of the annual arithmetic mean (AAM) NAAQS.  This control strategy developed
for attaining the 24-hour NAAQS should still be applied to the AAM averaging
time to assure attainment and maintenance of the AAM NAAQS.

6.1.1  24-Hour Maxima

     Because the short-term concentrations appear to be "controlling", these
concentrations were examined first.  Table 6.1 lists the relative source
contributions to the 1986 24-hour design value.  The table shows that the
main contributing sources are copper ore crushing and conveying, roadways,

                                    6.1

-------
                                 Table 6.1

                   Source Category Contributions (ug/m3)
                to  1986 Base Year  24-Hour  PM^Q Design Value^
Source Category
                                               Contributions to
                                             Ambient Concentrations
                                                    (ug/m^)
Copper Ore, Unloading, Crushing, Conveying

Road Dust (Excluding Kennecott and Canyon
 Roads)

Locomotives

Kennecott and Canyon Road Dust

Gypsum

All Others

Background

Total All Sources
                                                      410.8

                                                      100.5


                                                       58.8

                                                        5.3

                                                       10.2

                                                       16.2

                                                       16.0

                                                      618
iGarfield monitor location of point highest second high 24-hour
 concentrations.
                                    6.2

-------
and locomotive exhaust.  The total required reduction necessary  at Hayden
to meet the 24-hour NAAQS based on the 1986 design value is as follows (the
24-hour concentration shown here and in the remainder of this document refer
to the highest second high 24-hour value):

     TR = 618 - 150 = 468 ug/m3
     (150 ug/m3 is the 24-hour Average PM^o NAAQS)
This indication of needed reduction, however, neglects the growth potential
during the period between the 1986 design value and the 1991 attainment
date, and neglects also the potential impacts of "banked emissions", i.e.
those permitted sources not operating or operating at less than a normal
rate.  Growth and banked emissions must be considered for both 1990
attainment and for NAAQS maintenance through 1997.

6.1.2  Annual Arithmetic Mean

     Table 6.2 lists the relative source concentrations to the 1986 AAM
design value.  The total required reduction to meet the 1986 AAM NAAQS is as
follows :

     TR » 102.5 - 50 = 52.5 ug/m3
     (50 ug/m3 is the AAM NAAQS)

The influence of growth and banked emissions for 1990 and 1997 will, in the
case of Hayden, increase the reductions required for NAAQS attainment and
maintenance .

6.2  Growth Prelections and Banked Emissions

     Few guidelines exist for estimating growth and its air quality impacts.
For the Hayden. area, contact with city officials and the Pinal-Gila Counties
Air Quality Control District indicated near-term growth in the Hayden area
is entirely dependent on the copper industry; the officials did not foresee
growth in the area from retirees seeking a warm climate as is the case in
other Arizona areas.  The possibility of this type of growth, however,
should still be considered for the longer term.

     The ASARCO smelter operation is not expected to significantly increase
production because of limited equipment capacity and restrictions on sulfur
dioxide emissions.  The Kennecott smelter (now owned by ASARCO), although
not operated since before 1980, still has valid permits and therefore
potential banked emissions.  Resumption of Kennecott smelter operation would
cause increases in ore crushing and conveying, in locomotive traffic, and in
traffic on both paved and unpaved streets and roads.  It was assumed that
the Kennecott smelter when operating would share the current ore transport,
unloading, crushing, and conveying system.  The 1990 projections include the
banked Kennecott emissions and a 20 percent increase in locomotive use and
overall vehicle traffic from the increased activity and employment.  Table
6.3 lists the source contributions and highest 24-hour concentrations for
1990.  Table 6.4 lists the maximum AAM concentrations for 1990.
                                    6.3

-------
                                 Table 6.2

                   Source Category Contributions (ug/ra3)
                       to 1986 PMio AAM Design  Value1
                                               Contributions to
                                             Ambient Concentrations
Source Category                                     (ug/m^)
Copper Ore, Unloading, Crushing, Conveying             61.7

Road Dust (Excluding Kennecott and Canyon              11.0
 Roads)

Locomotives                                             5.4

Kennecott and Canyon Road Dust                          2.1

Gypsum                                                  1.9

All Others                                              4.4

Background                                             16.0

Total (All Sources)                                   102.5



^Garfield monitor location of point of maximum AAM.
                                    6.4

-------
                                 Table 6.3

                   Source Category Contributions (ug/m^)
                       to 1990 24-hour Maximum
                              Concentrations1
                                               Contributions to
                                              24-Hour Second High
                                                Concentrations
Source Category                                     (ug/m^)
Copper Ore, Unloading, Crushing,  Conveying            493

Unpaved Road Dust (Excluding Kennecott and             88
     Canyon Roads)

Paved Road Dust                                        32.6

Locomotives                                            70.6

Kennecott and Canyon Road Dust                          6.4

Gypsum                                                 10.2

All Others    .                                         18.5

Background                                             16.0

Total (All Sources)                                   735
     24-hour maximum refers to the highest second high concentration in the
 non-attainment area.
                                    6.5

-------
                                 Table 6.4

                   Source Category Contributions (ug/m3)
                        to 1990 PMin Concentrations1
                                               Contributions to
                                             Ambient Concentrations
Source Category                                     (ug/m^)
Copper Ore, Unloading, Crushing, Conveying             74

Unpaved Road Dust (Excluding Kennecott and              9.6
 Canyon Roads)

Paved Roads                                             3.6

Locomotives                                             6.5

Kennecott and Canyon Road Dust                          2.5

Gypsum                                                  1.9

All Others                                              4.7
              »
Background                                             16.0

Total (All Sources)                                   119



^Garfield monitor is location of point of maximum AAM.
                                    6.6

-------
     For 1997,  it was estimated that smelter-related sources would remain
unchanged but that vehicle traffic would increase by 20 percent over 1990
levels and that the "all other" and "background categories" would increase
by 10 percent over 1990 levels.  The projections of 1997 air quality must
consider the results of the control strategy implemented to meet the NAAQS
in 1990.  The control strategy becomes an iterative procedure, with
selection of a 1990 control strategy application of 1997 projections, and
adjustments to the 1990 control strategy to maintain the NAAQS through 1997.
An alternative would be implementation of further controls between 1990 and
1997.  The projected air quality for 1990 will be shown in Section 7 based
upon the 1990 air quality resulting from implementation of the control
strategy.

6.3  Emission Reductions Needed for 1990 NAAQS Compliance

     The modeled 1990 Hayden area air quality with the banked Kennecott
smelter emissions and no additional controls beyond 1986 levels showed the
area of maximum impact still to be near the Garfield monitor.  The largest
impacting sources were the ore unloading, crushing, and conveying
operations.  Neither the Kennecott smelter nor the ASARCO smelter
pyrometalurgical activities had significant impacts on maximum PM^o
concentrations.  Because the maxima occurred on calm days, the tailings pile
emissions also were not evident.  Separate modeling for days of high winds
still did not show exceedances resulting from tailings pile emissions.

6.3.1  Needed Reductions for 1990 NAAQS Attainment

     Tables 6.3 and 6.4 list the 1990 projected 24-hour and annual average
PM^Q concentrations before control strategy implementation.  The total
reductions of 1990 PM^o levels needed to attain the NAAQS are given below.
     0    24-hour maximum

               TR (ug/m3) = 735 - 150 - 585 ug/m3

     0    Annual Arithmetic Mean

               TR (ug/m3) = 119 - 50 = 69 ug/m3

6.4  Source Specific Control Options

     For control strategy development the control options for each source
category should be tabulated, evaluated, and ranked.  The ranking should be
based on technical feasibility and cost, and ultimately on the relative
contribution of each source to ambient concentrations.  For Hayden, Table
6.5 lists the options available for emission reductions from each source
category, including smelter emissions and tailings pile.  Although neither
smelter nor the tailings piles significantly impact maximum concentrations,
the source apportionment results indicate that they contribute
disproportionally to toxic air contaminant concentrations.  The following
subsections discuss the control options.


                                    6.7

-------
                                 Table 6.5

                      Source Specific Control Options
                         for Hayden  PM^Q  Attainment
Source Category and Control Alternative
 Additional
  Specific
Source Control
Copper Ore Dust Sources

      a.  Replace existing type W Rotoclones                90
          controlling crusher and transfer point
          emissions with fabric filter collectors.

      b.  Increase enclosure of unloading                   90
          operation and transfer points.

      c.  Combination of above for Source Category.         90

Unpaved Roads, Alleys and Parking Lots

     a.   Paving of most heavily travelled streets,         90
          i.e. Kennecott Road and Canyon Drive.

     b.   Paving of all other roads and alleys.             90

     c.   Combination of above for Source Category.         90

Paved Roads

     a.   Adding curbs to major paved streets.              50

Locomotives

     a.   Fuel injector adjustments to limit                60
          smoke generation.

      b.  Addition of collection devices.                   70
Gypsum
          Enclosure and vent to fabric filter
          collector.

                                    6.8
      80

-------
                             Table 6.5 (cont.)
Source Category and Control Alternative
                                                  Additional
                                                   Specific
                                                 Source Control
Smelter Furnace Building - ASARCO

     a.   Further expand flash furnace and
          converter hooding systems.

     b.   Add control system or process
          modifications to anode furnaces.

     c.   Combination of above for Source Category.

Smelter Furnace Building - Kennecott

     a.   Install secondary hooding on converter.

     b.   Add control system or process
          modifications to anode furnaces.

     c.   Install additional reverberatory
          furnace fugitive emission hooding
          system.

     d.   Combination of above for Source Category.

Tailings Pile

     a.
     b.
     c.
     Application of soil stabilization
     chemicals.

     Introduce vegetation with necessary
     soil supplements and irrigation.
     Install wind fences.

d.   Combination of above for Source Category.
50


80


70



90

80


50



75



80


90


40

90
                                    6.9

-------
6.4.1 Copper Ore Dust

     Copper ore dust emits from rail car unloading, ore conveying to the
crusher, the crushing operation, and transfer operations from the crusher
area to the concentrator area.  The unloading area, the crusher, and most of
the transfer operations are enclosed or hooded, and exhausted to wet
scrubber dust collectors.  Currently an estimated 45 percent of total copper
ore dust emitted exhaust through the wet scrubber stacks.  Uncollected
fugitive emissions comprise the remaining 55 percent.  Increased dust
collector efficiencies, and better dust capture through improved hooding and
enclosures provide the options for control of copper ore dust.  Replacement
of the existing American Air Filter Type W Rotoclones with fabric filter
collectors and improved hooding and enclosure of transfer points and venting
to fabric filter collectors is expected to reduce these emissions overall by
90 percent.

6.4.2  Un paved Roads, Alleys, and Parking Lots

     Paving of unpaved traffic ways are generally expected to reduce dust
emissions from these sources by 90 percent.  For Hay den, the control options
are (a) to pave the heavily traveled unpaved streets of Kennecott Road and
Canyon Drive that currently represent 2 percent of the total category impact
and (b) to pave the remainder of the unpaved roads, alleys, and parking lots
in the Hayden area.  Dust suppression measures other than paving were
considered unpractical for control of these sources.  Paving of Kennecott
Road and Canyon Drive and paving of other roads, streets and alleys would
reduce these total emissions by 90 percent.

6.4.3  Paved Roads

     Adding curbs to the most heavily travelled streets, and occasional
sweeping and washing would reduce emissions from paved streets by
approximately 50 percent.

6.4.4  Locomotives

     More frequent locomotive engine maintenance and adjustment and setting
of fuel injectors to limit smoke would reduce PMjo emissions an estimated 60
percent.

6.4.5  Gypsum Loading

     The current gypsum loading process has no controls; the trucks dump in
on open area into a hopper, and the hopper dumps into open railcars.
Enclosure and ventilation of this operation to a fabric filter collector
would reduce PM^g emissions an estimated 90 percent.
6.4.6  Smelter Furnace Building - ASARCO

     ASARCO furnace building emissions were divided into two sources (a)
those emissions escaping the hooding during tapping and slag skimming of the
flash furnace and charging, tapping and slag skimming of the converter, and

                                    6.10

-------
(b) emissions from the anode furnace.   Recent modifications, i.e., the
replacement of reverberatory furnaces  with a flash furnace, and adding
secondary hooding to the converters,  reduces the potential for further
reduction through additional hooding to only 40 percent of existing emission
rates, and therefore only 20 percent of total category emissions.  Control
of the anode furnace through additional control systems or process
modification could reduce emissions by 80 percent from all furnace, and 30
percent overall.

6.4.7  Smelter Furnace Building - Kennecott

     The Kennecott smelter has no secondary hooding on the converter, and
therefore the converter contributes 60 percent of the current furnace
building emissions.  Anode furnace emissions contribute 25 percent of the
total, and reverberatory furnace fugitives 15 percent of the total.  Adding
secondary hooding to the converter would achieve the largest reduction, with
control of the anode furnace and improved reverberatory fume hooding
providing less reduction.

6.4.8  Tailings Pile

     All control options for tailings piles pertain to windblown dust.
These controls include (a) application of soil stabilization chemicals,
introduction of vegetation with the necessary soil supplements and
irrigation, and the installation of wind fences.  Table 6.5 lists the
estimated percentage emission reductions for each of these control options.
                                    6.11

-------
7.0  SIP CONTROL STRATEGY IMPLEMENTATION
     As suggested in the PM^Q Guideline Document,  states should, to the
extent possible, utilize the existing TSP control strategy in the SIP as the
basis for a PM^g program sufficient to attain and maintain PM^g NAAQS.
Depending on the circumstances,  the control strategy can contain either
total particulate matter emission limits, or PM^g specific emission limits,
or a combination of both.  The finalized source emission rates developed
through receptor modeling and dispersion model reconciliation pertain
directly to PM^g emissions, not  total particulate emissions.  States that
are developing SIP emission limitations should assure that the regulations
either address PM^g emissions directly or set total emission limitations at
levels sufficient to achieve equivalent results.

     This section will describe  the process of selecting and testing the
control strategy and designing the strategy to maintain the NAAQS through
1997.  It will also discuss the  implementation process and implementation
scheduling .

7.1  Control Strategy Selection  Methodology

     Control strategy selection  involves choosing various control options
for each of the major contributors, such as those listed in Table 6.5, and
evaluating them through modeling to determine the effectiveness in attaining
the NAAQS.  Section 6 of the PM^g Guideline Document cautions that high
concentration locations within the non-attainment can be heavily related to
a single source, and strategies  to attain the NAAQS at a single location may
not be adequate for other locations.  Selection of the strategy should
therefore consider more locations than the design location for evaluating
emission reductions.

     The 1990 control strategy selection must relate to the need for
maintaining the NAAQS through 1997.  The procedure is to tentatively select
a 1990 control strategy and then apply growth projections to the controlled
sources to project 1997 air quality.  If 1997 projections  indicate
exceedances of the NAAQS, either the 1990 control strategy must be revised,
or provisions made for further tightening of control after 1990.

7.2  Implementation Methodology

     Emission regulations are commonly specified in terms of maximum
allowable pollutant concentrations, mass per unit of output or input, mass
per unit time, opacity or specified control efficiencies.  Regulating
emissions from fugitive sources  is more complicated and may involve rules
specifying equipment or operational procedures, or for specific sources,
negotiated agreements for source owners to install additional hooding, pave
roads and add curbing, and apply dust suppression chemicals, etc.  The Clean
Air Act provides authority to enforce the control measures necessary to
achieve attainment.

     The Clean Air Act requires  that the states present PM^g SIPs to EPA
within nine months of NAAQS promulgation.  A six month period is allowed for

                                    7.1

-------
EPA SIP approval, followed by a three year period to achieve NAAQS
attainment.

7.3  Hayden PMm Control Strategy and Implementation
     EPA promulgated the PM^Q NAAQS in July 1987, with the effective date of
July 31, 1987.  The affected states must submit SIP's by April 30, 1988, and
EPA must approve plans within 6 months, which is October 31, 1988.  The
three years for full SIP implementation and NAAQS attainment extend to
October 31, 1991.

     Following is the combination of control measures selected as a control
strategy to achieve the NAAQS in Hayden by 1990 and to maintain the NAAQS
through the year 1997.  Tables 7.1 and 7.2 list the modeled results of
control strategy implementation for the 1990 NAAQS attainment date and
demonstrate attainment.  Figure 7.1 shows the implementation schedule.

Control Measure 1 - Control Copper Ore Dust

     Replace current scrubbers with fabric filter collectors and increase
     enclosure, hooding, and venting to dust collectors of the fugitive
     emission points for an overall estimated reduction of 90 percent.

     The replacement with fabric filters will be achieved by modifying
     regulations to specify total remivaT. efficiencies, or further
     r-i.-5 trie ting the allowable mass per unit volume rate.  An alternative
     would be a negotiated agreement with ASARCO to upgrade or replace the
     existing equipment to meet these criteria.  Further enclosure and
     venting to control systems of fugitive sources will implemented by
     either applying current fugitive dust rules, further tightening of
     these rules, or agreements with ASARCO for specific modifications.
              •
     Rule development and procedures for control measure implementation
     through negotiated agreement will commence in August, 1988, two months
     prior to the Plan approval date, and will continue through March of
     1989.  Pre-engineering work on control systems will commence prior to
     final negotiations, with installation and modifications completed by
     March 1990.

Control Measure 2 - Paving of Unpaved Streets,  Roads and Alleys

     The paving of all significant unpaved roads and streets, including
     Kennecott Road and Canyon Drive would reduce emissions from these
     sources an estimated 90 percent.  Kennecott Road in Hayden, although
     used by the public, is currently owned by ASARCO, was originally a
     Kennecott smelter haul road.  Agreements with ASARCO for the paving of
     this road would implement this strategy.  Paving of Canyon Drive and
     other unpaved roads, streets, and alleys in Hayden would require
     agreements with the town of Hayden.

     Investigation into funding for the paving projects will commence in
     June 1988, with possible resolution by February 1989.  Specifications,

                                    7.2

-------
                                           Table  7.1

                          1990 Source Category Contribution  and Control
                                Strategy  for 24-Hour Maximum(l)
                                         NAAQS Demonstration

Source Category
Copper Ore, Unloading,
Crushing, Conveying
Unpaved Roads (Excluding
Kennecott and Canyon Roads)
Paved Roads
Locomotives
Kennecott and Canyon Roads
Gypsum
All Others
Background
Total All Sources
NAAQS
Projected 1990
24-Hour Maximum
Concentrations
(ug/m3)
493
88.0
32.6
70.6
6.4
10.2
18.5
16.0
735.3

Projected 199
24-Hour Conce
Source trations wi
Category Percent Control Strat
Reduction (ug/m3) Control (ug/m3)
<444> 90 49
<79.2> 90 8.8
<16.3> 50 16.3
<42.4> 60 28.2
<5.8> 90 0.6
<9.2> 90 1.0
— — 18.5
16.0
<596.9> 138.4
150
O)24-hour  maximum refers  to  highest  second high concentrations.
                                            7.3

-------
                  Table 7.2

1990 Source Category Contribution and Control
  Strategy for AAM PM^Q NAAQS Demonstration
Source Category
Copper Ore, Unloading,
Crushing, Conveying
Unpaved Roads (Excluding
Kennecott and Canyon Roads)
Paved Roads
Locomotives
Kennecott and Canyon Roads
Gypsum
All Others
Background
Total All Sources
NAAQS
Contribution to
1990 AAM (ug/m3)
Uncontrolled
74
9.6
3.6
6.5
2.5
1.9
4.7
16.0
119

Contribution
Source Category Percent 1990 AAM (ug/i
Reduction (ug/m^) Control Controlled
<66.6> 90 7.4
<8.6> 90 1.0
<1.8> 50 1.8
<3.9> 60 2.6
<2.2> 90 0.3
<1.7> 90 0.2
— 4.7
— 16.0
<84.8> 34.0
50
                  7.4

-------
                                          FIGURE 7.1




                       SIP PM10 IMPLEMENTATION SCHEDULE FOR HAYDEN, ARIZONA

PMjQ NAAQS Promulgations
SIP PM10 Submittal
SIP Plan Approval
Copper Ore Emission
Control Systems
Rules & Negotiations
Engineering & Installation
Road Paving
Negotiations & Finance
Design & Construction
Curbs & Gutters
Negotiations & Finance
Design & Construction
Locomotives
Rules & Negotiations
Engine Modifications
Gypsum Source
Rules & Negotiations
Engineering & Installation
Required Attainment Date
1987 .
7
1988
4
	
	
A 	 	
8 	
Q 	

1989
10
9 	
•2
9 	
-2
9 	
—3

9 	

1990
	 ^
	 „ *;
. 	 	 s


1991
7
Note: Numbers refer to months.

-------
     contract award, and paving will require an estimated 15 months, with
     paving complete by May 1990.

Control Measure 3 - Adding Curbing to Major Paved Streets

     Adding curbs and gutters to main streets limit soil intrusion, will
     reduce street dust loading and thus emissions an estimated 50 percent.

     Implementation of this measure will require agreements with the town of
     Hayden for street upgrading.

     The project will be concurrent with Control Measure 2.

Control Measure 4 - Locomotive Emission Reduction

     Modification and maintenance of locomotive engines in accordance with
     established diesel engine particulate emission control procedures would
     reduce emissions from the source category an approximate 60 percent.

     Negotiations with the railroad to modify the locomotive fleet and
     require periodic maintenance, and development and enforcement of
     visible smoke emission limits would implement this control measure.

     Development of an approach to implementation of this measure will
     commence in August 1988, with a regulation or negotiated agreement
     completed by the end of March 1989.  Full implementation will require
     approximately 9 months, with completion by the end of December 1989.

Control Measure 5 - Control of Gypsum Dust Sources

     Control systems will reduce this emission an estimated 90 percent.

     Negotiations with source owners to enclose the unloading area and
     install controls will implement this procedure.

     This measure will follow the same schedule as the copper ore control
     system, with rules and negotiations conducted from August 1988 through
     May 1989, and design and construction occurring from February 1989
     through March 1990.

7 .4  Preconstruction Review

     New major sources or modifications to major sources in Arizona are
subject to state requirements for preconstruction review under Arizona's
     PSD (Prevention of Significant Deterioration ) permitting program.
Note:  The PMjQ PSD requirements for Arizona were not available for this
report.  EPA could transfer the authority for PM^o PSD review to Arizona
upon approval of a PSD program which meets the requirements of the Part 52
PSD regulations.
                                    7.6

-------
7.5  Demonstration of NAAQS  Maintenance through 1997

     Tables 7.3 and 7.4 show the projected emissions growth for 1997, which
was estimated as a result of an increase in emissions from traffic (20
percent),  "other sources" (10 percent)  and the background concentration.
All other  sources remained at 1990 levels.  The emissions increase was based
on 1990 air quality following control strategy implementation.  The tables
demonstrate that the control strategy is sufficient to maintain the NAAQS
through 1997.

7.6  Annual Progress Report

     On an annual basis, the State will evaluate and demonstrate their
progress in reaching the goal of attainment and maintenance of the PM^g
NAAQS through 1997.  This will be achieved by the State by providing the
following  information in the form of an Annual Progress Report.

     The first Annual Progress Report is due to EPA, Region 9, six months
after the  first full calender year of implementation.  The subsequent annual
reports are due six months after the end of the calender year.

     For each control measure the State has committed to implement in the
     SIP,  the annual report  will include:

     a)   the names of each  source affected;

     b)   the current status of implementation of the control measure;
          This will include  summarized information on the regulation
          development (if regulation revision was necessary), negotiation
          with the sources and/or municipalities affected, project
          financing, engineering design and/or construction progress for the
          year, as appropriate.

     c)   planned versus actual implementation and/or adoption dates of the
          control measure and/or regulation;

     d)   planned versus actual effectiveness of the control measure;

     e)   planned versus actual decrease/increase in contribution to ambient
          concentration;

     f)   planned versus actual effect that area growth has had on the
          emissions reductions;

     g)   short explanation  of shortfalls, where applicable;
          If shortfalls have occurred,  the State will include a plan to
          develop a contingency plan or SIP revision which would be
          necessary to attain or maintain the NAAQS.
                                    7.7

-------
                                 Table 7.3

               1997 Source Category Contributions Reflecting
                      Projections and Control Strategy
                    24-Hour PM^Q Maximum Concentrations
                                  (ug/m3)
                                             Contributions to
Source Categpry                         Ambient Concentrations (ug/m3)
Copper Ore, Unloading, Crushing                    49
 Conveying

Road Dust (Excluding Kennecott and                 10.6
 Canyon Roads)

Paved Roads                                        19.6

Locomotives                                        28.2

Kennecott and Canyon Road Dust                      0.7

Gypsum                                              1

All Others                                         20.3

Background                                         17.6

Total All Sources                                 147

NAAQS                                             150


Note:  Maximum 24-hour means highest second high 24-hour concentration.
                                    7.8

-------
                                 Table 7.4

                     1997 Source Category  Contributions
                Reflecting Projections and Control Strategy
                         (AAM PM^Q Concentrations)
                                  (ug/m3)
                                             Contributions to
Source Category                         Ambient Concentrations (ug/m3)
Copper Ore, Unloading, Crushing                     7.4
 Conveying

Road Dust (Excluding Kennecott and                  1.2
 Canyon Roads)

Paved Roads                                         2.2

Locomotives                                         2.6

Kennecott and Canyon Road Dust                      1.3

Gypsum                                              1

All Others                                          5.2

Background                                         17.6

Total All Sources                                  38.5

NAAQS                                              50
                                    7.9

-------
8.0  AIR TOXICS SOURCES AND EMISSIONS

     Toxic air contaminants (TACs) can comprise a portion of ambient
concentrations.  Therefore, the reduction of TAG emissions, and the adverse
health effects associated with them, can often be achieved through the
application of controls to specific sources of PM^Q-  The impact of the
proposed control strategy options on potential adverse health impacts from
TAG emissions should be evaluated during SIP development.

     For Hayden, carcinogenic and non-carcinogenic air toxics were
identified through the source characterization and ambient monitoring
programs performed as part of the SIP development process.  The impact of
the proposed.control strategy on health risks from these emissions was
examined.

8.1  Health Risk Assessment

     A quantitative assessment of health risks associated with TAG emissions
should be performed when significant TAG sources have been identified in the
SIP study area.  The evaluation should include an analysis of both chronic
(long-term) and acute (short-term) health effects.  Simple screening level
procedures may be sufficient in some cases while more elaborate formal risk
assessment techniques may be required for complex situations.

     Unfortunately, limited data and resources were available for a thorough
assessment of the health risks at Hayden.  As a result, a rough estimate of
the total individual excess cancer risk resulting from human inhalation of
existing ambient TAG levels was made.  Three known or suspected human
carcinogens (cadmium, chromium and nickel) were identified in samples
collected during the sampling program performed in late 1986 and early 1987.
It was assumed that the average concentrations of the substances collected
over this limiped period were indicative of annual-average concentrations in
order to estimate the individual excess cancer risks.  This analysis also
assumed that the chromium measured was 100 percent hexavalent chromium.
This is a conservative assumption since a portion of the measured values is
the non-toxic trivalent form.

     Individual excess cancer risks were calculated by multiplying the
annual-average concentrations of each substance with its associated unit
risk factor (URF).  A unit risk factor is the estimated probability of a
healthy individual contracting cancer as a result of a constant exposure to
an ambient concentration of 1 ug/m3 of a carcinogenic substance over a 70
year period.  The URFs used were established by the EPA and the California
Department of Health Services.

     Table 8.1 shows that the estimated total individual excess cancer risk
from exposure to measured TAG levels is approximately 1.5 incidences of
cancer per 10,000 people exposed.

     The ambient samples identified the following non-carcinogenic air
toxics; manganese, copper, zinc and arsenic.  A health risk assessment of
these compounds was not performed.

                                    8.1

-------
                                 Table 8.1

       Individual Excess Cancer Risk from Measured Air Toxics Levels


                      Measured             Unit        Excess Individual
                Concentration (ug/m^)      Risk           Cancer Risk	
Substance        Hayden      Garfield     Factor      Hayden      Garfield


Cadmium     .   5.4 x 10~3   5.2 x 1Q-3  1.8 x 10~3  9.72 x 10~°  9.36 x 10~°

Hexavalent     9.0 x 1Q-*   8.0 x 10"4  1.5 x 10~l  1.35 x ICT*  1.20 x ICH*
 Chromium

Nickel         2.8 x 10~3   5.1 x 10~3  5.9 x 10~4  1.65 x 10~6  3.01 x lQ-°

Total                                           .    1.46 x 10~4  1.32 x 10~4
                                    8.2

-------
8.2  Toxic Emission Source Identification

     The NEA,  Inc.  report included  as Appendix A lists the XRF analysis
results of source samples called  from all the pyrometallurgical sources and
from the crusher, roads,  etc.   For  lead,  arsenic and cadmium the NEA, Inc.
report shows in its Appendix A the  relative contribution of these sources to
the ambient air concentrations.   With minor exceptions, such as copper, the
pyrometallurgical sources such as the converter, flash furnace, and slag
skimming and pouring operation contributed 60 to 80 percent of the ambient
concentrations of the elements.   The control strategy selected for
not include reduction of  these source emissions due to their inherent
technical and emissions problems  compared to controlling crusher dust
emission and- paving roads.

8.3  Control Options for  Toxic Air  Contaminants
     Because the PM^o control strategy does not include the major toxic
sources, implementation of the control strategy will only marginally (less
than 20 percent) reduce the ambient concentration of the toxic pollutants
listed.

     Possible control measures for the sources within the smelters suspected
of emitting toxic air contaminants are discussed in Chapter 6 of this
document.  (Subsection 6.4 and Table 6.5).   Implementation of controls on
all of these sources with the exception of the anode furnaces, would reduce
TAC concentrations in the atmosphere an estimated 50 percent.

8.4  Further Studies

     Resources should be directed toward the development of a data base
which will allow for a more rigorous evaluation of the health risks
associated with TAC emissions in Hay den.  This would include a source
sampling program designed to identify all major TAC sources.  From this
program, an air toxics emission inventory would be generated and control
strategy effectiveness would be evaluated through advanced dispersion
modeling.  Both carcinogenic and non-cancer health risks for the exposed
population and workers at the smelters would be assessed for the 1986 base
year and future years.
                                    8.3

-------
9.0  REFERENCES

Compilation of Air Pollutant Emission Factors,  Fourth Edition and
Supplements, AP-42, U.S.  Environmental Protection Agency, Research Triangle
Park, NC, September 1985.

Compilation of Preliminary Particle-Sized Emission Factor Data Base, EPA-
450/4-82-016, U.S. Environmental Protection Agency, Research Triangle Park,
NC, November 1982.

Core, J.E., J.J. Shah,  and J.A.  Cooper, Receptor Model Source Composition
Library, U.S. EPA 450/4-85-002,  U.S.  Environmental Protection Agency,
Research Triangle Park, NC, November  1984.

Cowherd, C. Jr., and Kinsey, J.S.,  Identification, Assessment, and Control
of Fugitive Particulate Emissions,  EPA-600/8-86-023, August 1986.

Dzubay, T.G., and Harbour, R.K., A  Method to Improve the Adhesion of Aerosol
Particles on Teflon Filters, Journal  of Air Pollution Control Association,
V. 33, p. 392, 1983.

Engineering-Science, Air Toxics  Source Assessment Manual for California Air
Pollution Control Districts, for U.S. Environmental Protection Agency,
Region IX, San Francisco, CA, September 1986.

Engineering-Science, Development of an Emission Inventory for Urban Particle
Model Validation in the Philadelphia  AQCR. Contract No. 68-02-3509 with the
U.S. Environmental Protection Agency, Research Triangle Park, NC, 1984.

Guideline on Air Quality Models  (Revised). EPA-450/2-78-027R, July 1986.

Houck, J.E., Cooper, F.A., and Larson, E.R., Dilution Sampling for Chemical
Receptor Source Fingerprinting,  Proceedings of Air Pollution Control
Association, New Orleans, LA, June  1982.

Industrial Source Complex (ISC)  Dispersion Model User's Guide - Second
Edition - Volume 1, EPA-450/4-86-005a, June 1986.

Martin, N., Hayden City Clerk, Telecon Concerning Hayden Area Growth
Projections, Hayden, AZ,  July 1987.

NEA, Inc., Source Apportionment  of  Suspended Particles and Toxic Elements in
Hayden, Arizona, for EPA under Engineering-Science Contract, Beaverton, OR,
April 15, 1987.

Nickling, W.G. and Gillies, J.A., Evaluation of Aerosol Production Potential
of Type Surfaces in Arizona, MND Associates under EPA Contract through
Engineering-Science, July 1986.

Noonan, F.M., Inhalable Particulate Matter (PMm) Emission Factor Program -
Status Report, U.S. Environmental Protection Agency, Research Triangle Park,
NC, January  1985.

                                   9.1

-------
Pace, T.G. and Watson, J.G., Protocol for Applying and Validating the CMB
Model, U.S. Environmental Protection Agency, Research Triangle Park, NC.,
and Desert Research Institute, Reno, NV, January 1987.

Pace, T.G., Anderson, M. and DeCesar, R., Protocol for Reconciling
Differences Among Receptor and Dispersion Models, U.S. Environmental
Protection Agency, Research Triangle Park, NC, and TRC Environmental
Consultants, East Hartford, CT.  December 1986.

Patterson, R.G. et al., Controls for Fine Particulate Emissions from
Industrial Sources in California, for California Air Resources Board,
Contract No. A9-119-30, 1986.

Peterson, Arlo, Meetings and Telephone Conversations Concerning
Meteorological Data and Emissions Inventories, ASARCO, Hayden, AZ, 1986 and
1987.

PEI Associates, Inc., PMm and Fugitive Dust in the Southwest - Ambient
Impact, Sources, and Remedies, U.S. Environmental Protection Agency,
Contract No. 68-02-3512, Research Triangle Park, NC, July 1985.

PMm SIP Development Guideline, U.S. Environmental Protection Agency, OAQPS,
Research Triangle Park, NC, Draft, September, 1986.

Puget Sound Air Pollution Control Agency, The Concentration of Lead,
Arsenic, Mercury, Cadmium, SO?, and Suspended Sulfated Downwind form the
Tacoma Smelter, the Impact and Control Status, and Benefits from Reduction.
Seattle, WA, 1974.

Revised CMB User's Manual, U.S. Environmental Protection Agency, Research
Triangle Park, NC, January 1987.

Traffic on the Arizona Highway System, Arizona Department of Transportation,
1984.
                                    9.2

-------