&EFW
United States
Environmental Protection
Agency
Environmental Research
Laboratory
Duluth MN 558O4
EPA 600 3-80-007
January 1980
Research and Development
Sources and
Transports of
Coal in the
Duluth-Superior
Harbor
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency. have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to faster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. Special Reports
9. Miscellaneous Reports
This report has been assigned to the ECOLOGICAL RESEARCH series. This series
describes research on the effects of pollution on humans, plant and animal spe-
cies, and materials. Problems are assessed for their long- and short-term influ-
ences. Investigations include formation, transport, and pathway studies to deter-
mine the fate of pollutants and their effects. This work provides the technical basis
for setting standards to minimize undesirabfè changes in living organisms in the
aquatic, terrestrial, arid atmospheric environments. .
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/3-80-007
January 1980
SOURCES AND TRANSPORTS OF COAL IN THE DULUTH-SUPERIOR HARBOR
by
Michael Sydor
Kirby Stortz
Department of Physics
University of Minnesota, Duluth
Duluth, Minnesota 55812
Grant No. R-803952
Project Officer
Douglas W. Kuehl
Environmental Research Laboratory
Duluth, Minnesota 55804
ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
DULUTH, MINNESOTA 55804
-------
DISCLAIMER
This report has been reviewed by the Environmental Research Laboratory —
Duluth, U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of
trade names or commercial products constitute endorsement or recommendation
for use.
ii
-------
FOREWORD
As the dependence of the United States upon coal as a major source of
energy increases, so shall the problems associated with the mining, transport
and storage of coal. These problems include the adverse effects coal
particulates can have upon the ecosystem.
The Duluth—Superior Harbor on the western end of Lake Superior is
rapidly becoming a major shipping port for coal from the western states.
Concern for the water quality of the harbor has resulted in an effort to
evaluate the distribution of coal particulates in the harbor and subsequently
evaluate the effects of the coal particulates upon water quality.
This report describes the distribution of coal particulates throughout
the harbor as released from selected sites and measured by experimental data.
In addition the report describes a mathematical uQdel developed to predict
distribution of coal from spills.
J. David Yount
Deputy Director
Environmental Research Laboratory—Duluth
iLL
-------
ABS TRACT
Dispersion of particulates from ORTRAN coal transshipment facility was
investigated to estimate the input of coal dust into Duluth harbor and to
determine the transport of coal particulates to Lake Superior. A numerical
model was used to discuss dispersal of contaminants and determine the resi-
dence time of pollutants in the waterway. The model was verified using
measurements of water levels, currents, and water quality parameters.
The relative magnitudes of dust sources due to ship loading and coal
pile maintenance were obtained from particulate concentrations movements in
air and measurement of particle deposition rates. The deposition rates were
stablished from analysis of material trapped in collection buckets and
directional measurements of dust deposited on slides. The overall magni-
tudes of the dust sources as a function of winds were estimated using a
Gaussian plume model.
Resuspension by ship traffic acting as a secondary source of particu—
lates was investigated through measurement of suspended solids and turbidi—
ties in the resuspension plumes. The numerical model for the harbor was
subsequently used to estimate the transport of the resuspended material to
Lake Superior.
Remote sensing data correlated with measurements of dustfall and snow
reflectivity were used to identify the major dust sources in the harbor.
This report was submitted in fulfillment of Grant No. R—803952 by the
Environmental Research Laboratory — Duluth under sponsorship of the U.S.
Environmental Protection Agency. The report covers a period from July, 1975
to June, 1978, and was completed as of April, 1979.
iv
-------
Foreword
Abstract .
Figures
Tables
Acknowledgments .
iii
iv
vi
ix
K
References
80
CONTENTS
1. Introduction
2. Conclusions
3. Recommendations
4. Water Levels — Harbor Dynamics
5. Hydrodynamic Model
6. Water Quality Model
7. Windblown Coal Dust
8. Dispersal of Coal Dust Fallout in the Harbor
Waterway
9. Particulate Input During Ship Loading .
10. Resuspension Due to Ship Traffic
11. Pollutant Residence Time in the Harbor.
12. Snow Albedo
1
3
5
6
13
21
30
41
58
62
68
72
V
-------
FIGURES
Number Page
1 Duluth—Superior harbor . 7
2 Fourier analysis of water level data shows the long term
oscillation periods in the harbor and Lake Superior . . 11
3 Hydrodynamic model grid for the inner harbor 14
4 Calculated and measured currents up and down river from
the ORTRAN coal dock 17
5 Comparison of measured and calculated water levels 18
6 Daily variation in Duluth Sewage Plant discharge rate 26
7 Calculated and measured specific conductance at west Blatnik
Bridge junction 27
8 Distribution of coal dust fallout after 1 day of 6.8 rn/s
northeasterly winds 33
9 Distribution of wind directions for Duluth—Superior harbor
(4 year average) 35
10 Distribution of wind speeds for the Duluth—Superior harbor
(4 year average) 36
11 Estimated annual distribution of coal dust fallout based
on Gaussian plume model with constant source 37
12 Estimated annual distribution of coal dust based on Gaussian
plume model with source proportional to wind speed 38
13 Particle size distributions 42
14 Sampling stations for water quality parameters 43
15 Calculated settling rate for St. Louis River particulate 45
16 Estimated settling rate for red clay in Lake Superior shore
zone 46
vi
-------
Number Page
17 Stokes settling for St. Louis River particulate 48
18 Stokes settling for windblown coal dust from the pile and
at ship loading for northerly winds 49
19 Estimated Stokes settling for resuspended material 50
20 Settling of St. Louis River plume particulates in laboratory
column 51
21 Fallout from the coal pile after 2 days of 6.8 m/s
northeasterly winds 52
22 Sediment distribution from 2 day fallout during northeasterly
winds. The distribution was calculated for transports due to
15 cm seiche and no settling in high flow zones 53
23 Sediment distribution from 2 day fallout during northeasterly
winds. The distribution was calculated for 15 cm seiche and
settling in all channels 54
24 Sediment distribution from 2 days of fallout for northeasterly
winds. Distribution calculated for transports due to 3 cm
seiche 56
25 Sediment distribution resulting from yearly input of wind-
blown fines during ship loading — based on 15 cm seiche
transports 59
26 Sediment distribution resulting from yearly input of wind-
blown fines during ship loading — based on 3 cm seiche
transports 60
27 Estimated settling rate for resuspended material . . 63
28 Suspended solids track for coal carrier entering the harbor
through Duluth entry 65
29 Redistribution of sediment after resuspension from a ship’s
passage 66
30 Contaminant concentration at lake entries for a 200 kg
conservative input at coal dock (15 cm seiche conditions) . . . . 69
31 Contaminant concentration at lake entries for a 200 kg
conservative input at coal dock (3 cm seiche conditions). . . 70
32 Accumulated output at lake entries of a conservative
contaminant initially put in at the coal dock 71
vii
-------
Number
33 Relationship between fallout mass and cross—sectional area
for coal particulates deposited over snow 73
34 Relationship between snow albedo and percent of snow surface
covered with coal 74
35 Seasonal variation in Landsat CCT maximum reflectance for
Rice Lake 76
36 Snow albedo derived from 09JAN77 Landsat scene 77
37 Snow albedo derived from 27JAN77 Landsat scene 79
viii
-------
TABLES
Number Page
1 Geometrical and Flow Properties of the Duluth—Superior Harbor 8
2 Helmholtz Frequencies for Various Coupling of Inlets as a
System of Oscillators 9
3 Calculated Currents for Various Channels of Interest 19
4 Time Steps for Various Seiche Amplitudes Used in Water Quality
Model 24
5 Precipitation Recorded During Water Quality Simulation Period 28
6 Coal Particulate Fallout (102 gm/m 2 /day) 39
7 Kg of Coal Particulates Deposited in Lake Superior from a 574
Kg Fallout Over the Harbor 55
8 Transport of Particulates From the Coal Dock to Lake Superior 61
9 Percent of Total Load Resuspended Due to Ship Traffic Which
is Carried to Lake Superior 67
10 Coal Dust Deposition (i.ig/cm 2 /day) 75
lx
-------
ACKNOWLEDGMENTS
The work of Dr. Thomas F. Jordan, Physics Department, University of
Minnesota, Duluth, on harbor dynamics is gratefully acknowledged. We wish
to thank our co—workers Steve R. Diehi, Gordon J. Oman, David T. Smith, and
John A. Sorensen for their help in computations and reduction and analysis
of data.
We are indebted to Mr. Nelson Thomas, U.S. E.P.A. Grosse Ile, Michigan,
for some of the particle size analysis and to Mr. Thomas M. Biele, Lakewood
Filtration Plant, for use of Hiac PC 320 particle size counter.
Our thanks to Dr. Philip M. Cook, U.S. E.P.A., Duluth, for discussion of
particle size work and for advice and help on electron microscopy work. We
are indebted to our former colleague Mr. Duane W. Long, now with Western Lake
Superior Sanitary District, for chemical analysis of the samples.
Our sincere thanks to Mr. Courtland L. Mueller Jr. and Mr. Howard J.
Schwartz of the U.S. Army Corps of Engineers for information on the harbor
sedimentation and use of their facilities in field work.
x
-------
SECTION 1
INTRODUCT ION
This study is part of a continuing program designed to determine some
of the environmental effects resulting from the Nation’s increased usage of
coal. The purpose of this study was to investigate the input, dispersion,
and transport of coal contaminants within the Duluth—Superior harbor and
Lake Superior. Coal contaminants studied were those resulting from opera-
tion of the new Orba Rice Transshipment Company (ORTRAN) coal transshipment
facility located in Superior, Wisconsin.
To determine the characteristics of the particulate dispersal in the
harbor, it was necessary to predict mass transport in the entire waterway.
This required extensive measurements and the use of hydrodynamic and water
quality models. Similarly, description of the manner and magnitude of the
input of particulates into the harbor required measurements of particulate
concentration in the air and their deposition rates under various meteoro-
logical conditions. This data was extended through numerical modeling to
assess the total magnitude of dust sources.
McElroy and Chiu (1974) proposed a mathematical model for water quality
in the St. Louis River from Lake Superior to Brookston, Minnesota. Their
model was based on the Columbia River Model developed by Callaway, et al.
(1970) and documented by Feigner and Harris (1970). The Columbia River
Model has been used successfully in other investigations (Churchill 1976).
McElroy recommended that the 1974 St. Louis River Model be ammended to
include the effects of the Lake Superior seiche. The seiche is a complex
periodic oscillation of the lake’s water level caused by meteorological
factors such as winds and air pressure variations over the lake. Following
McElroy’s suggestion, a network of water level and current meters was
installed during 1975 and 1976 to measure seiche oscillations and determine
the effect of seiche on mass transport in the harbor. This data was neces-
sary for model verification. At the same time, the Columbia River Model was
modified to include the seiche. By the time ORTRAN became fully operational
in late 1976, the water quality model was extensively tested and verified.
However, in the short time since ORTRAN has been fully operational, it
has been impossible to determine accurately the distribution and seasonal
variation of dust sources as a function of maintenance practices, and all of
the various meteorological factors such as winds, precipitation, and
humidity. For this report, a few specific meteorological events were moni-
tored and modeled. These results were then extended using yearly wind
distributions and ship traffic data to roughly estimate the annual input of
1
-------
particulate into the harbor and to determine the transport of the particu—
late to Lake Superior. Although these annual estimates are crude, they do
provide reasonable estimates of the relative significance of various mechan-
isms for input of particulate into the harbor and lake.
2
-------
SECTION 2
CONCLUS IONS
Duluth—Superior harbor is a freshwater estuary. The dominant force
driving mass transport in the harbor is the Lake Superior seiche, which acts
on the harbor through two inlets. The seiche ranges in amplitude from 3 to
25 cm and reverses the flow in the St. Louis River up to Fond du Lac, the
site of the first river dam. For high seiche many of the main shipping
channels show oscillatory currents which exceed 20 cm/sec, a threshold
sufficient for resuspension of unconsolidated sediment.
Windblown dust from the ORTRAN coal transshipment facility is the major
source of coal particulates for the harbor. Some of this dust comes
directly from wind action on the coal pile, however, much of it is caused by
grooming of the pile by large caterpillar tractors. Generally, the wind-
blown material from ORTRAN deposits over land occupied by railways and other
industrial sites. However, for southerly and easterly winds dust from the
facility falls out over parts of the harbor, including the fish spawning and
recreational areas near the Arrowhead Bridge. During one day of 6.8 rn/sec
(15 mph) northeast wind, 287 kg of coal particulates were deposited in the
waterway southwest of the pile. Annually about 20 metric tons of coal
particulates are deposited in the harbor, and about 1.5 tons fall out
directly over Lake Superior. The most significant input of coal particulate
from the facility into the lake comes from the springtime transport to the
lake of about 4 tons coal dust which accumulates in the harbor ice cover.
The ship loading operation is a major source of coarse particulates.
During this process, coal is moved via a conveyor system from a large out-
door storage pile to large lake carriers, which transport the coal to
Detroit. Depending on wind conditions, between 20 and 260 kg of coal parti—
culates are deposited in the channel where the ships are tied up during
loading. The dust originates from the loading chute and the associated
conveyor system. This material is further dispersed by resuspension due to
ship traffic.
Concentrations of coal particulates in the harbor waterway range from
1 mg/9 in the channels near the loading dock to 1 pg/P.. in the channel where
chlorinated effluents are put out by the Duluth sewage plant. Generally
the concentration of coal particulate in the shipping channels ranges
between 1 pg/P.. and 20 pg/P.., roughly 0.1% of the average concentration of
suspended solids in the harbor. Less than 1% of the coal input into the
harbor water (mostly the fine particulates between 1 and 6 ii in diameter)
is transported to Lake Superior.. This amounts to an annual input of 50 to
3
-------
250 kg of coal into the lake. This input is small when compared to the
direct fallout of particulate over the lake.
A momentary spill of dissolved pollutant distributed uniformly through-
out the channel adjacent to the coal dock would take from 8 to 21 days to
reach Lake Superior at peak concentration. Depending on seiche amplitude,
the peak concentration of pollutant at the lake entries would range from
0.05 to 0.1% of the initial concentration in the loading channel. kt 0.1%
concentration level, the spill would extend over 30% of the harbor. The
contaminant would reach areas of the harbor extending from the Burlington
Railway Bridge, which is 2.5 km upriver from the coal dock, to the lake
entries, and would remain in the harbor at concentration levels above 0.01%
for 30 to 40 days.
Sedimentation of ORTRAI4 coal particulates in the harbor affects
primarily areas downriver from the Burlington Northern Railway Bridge.
Sedimentation levels of coal for one year range from 0.2 g/rn 2 for areas 3 km
from the source to 100 g/m 2 near the coal dock.
Resuspension of bottom sediments by ship traffic is an important
secondary source of harbor turbidity. Suspended solids in ship resuspension
plumes range from 10 to 50 mg/i, five times the usual concentration of
suspended solids in the harbor. An estimated i0 5 kg of material is resus-
pended per passage of a coal ship. The resuspended material is coarse and
settles rapidly, thus the output of resuspended material into Lake Superior
is negligible. Most of the material resuspended by ship traffic is redis-
tributed to the low turbulence areas within the transit channels.
During the winter, remote sensing data can be used to identify major
dustfall sources in the harbor. Remote sensing data coupled with snow
albedo measurements indicate that the input of dust from the C. Reiss Inland
Dock and the Hallet Storage Facility may exceed the input from the ORTRAN
coal transshipment facility. This is largely due to the prevailing wind
directions, and various maintenance practices at the bulk material storage
piles. The proximity of these old coal piles to the waterline is also a
factor.
4
-------
SECTION 3
RECOMMENDATIONS
Although the ORTRAN coal transshipment facility is located so as to
reduce the input of coal into the harbor and lake and appears relatively
clean in comparison to other storage facilities, the terminal is still a
substantial source of coal dust. A more detailed investigation of this
facility should be conducted to determine accurately the input of coal
particulates into the harbor and lake as a function of climatological factors
such as temperature, humidity, and winds. The effects of grooming of the
coal pile by large caterpillar tractors should be examined closely. In
particular the design of the caterpillar fans should be changed to minimize
the blowing of coal particles into air. The effectiveness of current dust
abatement procedures such as spraying of the pile should be examined through-
out the seasons and an effort should be made to determine what additional
measures could help diminish the dust source.
If the input of coal dust into the harbor and the lake is deemed signif-
icant in terms of toxic pollutants, t1- prevailing winds and the proximity of
the populated areas should be considered as first priority when locations of
future coal facilities are planned. To reduce the concentration of coal
particulates in the harbor and the lake, future coal shipping facilities
should be located on main channels, preferably in the inner harbor. Presence
of adjoining slips should be avoided, and the coal piles should be placed as
far from the water line as possible.
The older coal storage facilities common in many Great Lakes’ harbors
should be examined as sources of coal dust. Because these old facilities
are often located at the water line and generally do not have dust abatement
facilities, they are significant sources of coal particulates in the harbors.
Use of remote sensing is feasible as a means of identification and surveil-
lance of dust sources from bulk storage piles.
5
-------
SECTION 4
WATER LEVELS - HARBOR DYNANICS
To examine how water moves in the harbor, the forcing terms and the
geometry of the harbor need to be examined from the standpoint of wave propa-
gation and resonance characteristics. A bay or a harbor on the Great Lakes
acts like an estuary where water movement is driven by lake level oscilla-
tions and the harbor geometry (Freeman, Hatnblin, and Murty 1974). For
instance, a harbor Inlet can be considered as an oscillating plug of water
coupling the harbor (or bay) with the lake (Defant 1961, Miles 1974, Ippen
and Goday 1963). The geometry of the inlet and the bay reservoir could
select certain frequencies for mass movement, thus acting as a kind of
mechanical filter (Hwang and Tuck 1970, Carrier et al. 1971). A single very
small’ inlet would select only the long period water level oscillations
(Carrier et al. 1971). On the other hand, if more than one inlet is present,
the inlets could act in unison as coupled oscillators selecting certain
forcing terms and causing anomalous water transports in parts of the harbor.
These anomalous transports could be significant in resuspension of sediment
and in dispersal of pollutants (Bennett 1974).
The Duluth—Superior harbor is typical of Great Lakes harbors. The
waterway forms the mouth of the St. Louis River (Figure 1). It Includes some
50 km of shipping channels, whose depth generally ranges from 5 to 10 meters
with an average depth of 8.5 meters. The harbor Is subject to a lake seiche
which usually ranges in amplitude between 3 and 15 cm, but may exceed 25 cm
during severe storms. Table 1 shows the pertinent geometrical properties of
the harbor and inlets, and some flow characteristics derived from numerical
modeling and physical measurements.
To examine the harbor for resonance frequencies, note that there are two
lake entries or inlets to the Duluth—Superior harbor. There is also one
major harbor constriction under the Blatnik Bridge which separates the inner
harbor from the outer harbor. If these inlets and the constriction are
considered as a system of coupled oscillators, the following solution is
found for the angular frequency representing the Helmholtz resonances for
the harbor:
6
-------
MINNESOTA N
DULUTH
Drills
Figure 1. Duluth—Superior harbor. Water level stations arc
indicated by numbers, current meter stations by letters.
Entry LAKE SUPERIOR
South Entry
SUPERIOR
WISONSIN
9
10
kilometers
-------
2 1 ( 1+2+3+ S 3
2 A L 1 L 2 L 3 ±
S S 2 3 3)2 -
(_i + —
L 1 L 2 L 3 A 3
S
AL 3 A L 1
where A = area of outer harbor, A = area of the inner harbor,
S 1 , S 3 , L 1 , L 2 , L 3 are the cross—sectional areas and lengths
of the north entry, south entry, and inner harbor entry respec-
tively, and g = 9.8 rn/sec 2 .
TABLE 1. GEOMETRICAL AND FLOW PROPERTIES OF THE DULUTH—SUPERIOR HARBOR
12
L4A
(1)
Total Surface Area
Outer Harbor Surface Area (A) (east of Blatnik Bridge)
Inner Harbor Surface Area (A ) (west of Blatnik Bridge)
Surface Area in Shipping Channels
Total Volume
Volume in Shipping Channels
North Entry Cross—sectional Area (S 1 )
South Entry Cross—sectional Area CS 2 )
Inner Harbor Inlet Cross—sectional Area (S 3 )
North Inlet Length (L 1 )
South Inlet Length (L 2 )
Inner Harbor Inlet Length (L 3 )
% Flow in Dredged Channels
Discharge at North Entry (% of total)
Discharge at South Entry (% of total)
% of Total Harbor Volume Exchanged Per Day for
Average Seiche
% of Total Harbor Volume From Daily Outflow of
St. Louis River
Average Discharge Rate of St. Louis River
45 x ü 6 rn 2
16 x io 6 m 2
29 x io6 m 2
9.6 1o 6 m
0.17 x rn 3
.073 x io rn
796 m 2
1300 m 2
972 in 2
480 m
900 in
126 in
807
607
4 0/
6Z
2%
U in isee
8
-------
Table 2 gives the periods of Helmholtz frequencies for various assump—
TABLE 2. HELMHOLTZ FREQUENCIES FOR VARIOUS COUPLING
OF INLETS AS A SYSTEM OF OSCILLATORS
Coupling Scheme
Period In Hours
Three inlets acting as a coupled
system of oscillators
2.3
The north and south
inlets acting as a 6 2
A. 45x10 m
coupled oscillator bay
2.1
Single inlet acting as
an oscillator connecting
the bay and the lake
North inlet A. = 45 x 106 m 2
bay
South inlet A. 45 x 106 m 2
bay
2.9
3.1
Single inlet mode
North inlet A. = 16 x 106 m 2
bay
South inlet A.b = 16 x 106 m 2
ay
1.7
1.9
tions on the effective harbor area, Abay and oscillator coupling schemes.
It takes on the order of 30 minutes for a disturbance to propagate between
the entries. Periods shorter than 1 hour are damped out. Thus, the impor-
tant Helmholtz oscillation periods range from 1 hour to 3 hours.
Spectral analysis of the daily water level measurements shows oscilla-
tion peaks at 1.5, 1.9, 2.3, 2.8, 3.3, and 3.8 hours at the Allouez Bay
Station near the south entry to the lake, and 1.9, 2.3, 2.8, 3.3, and 3.8
hours at the Drills Marina Station well inside the inner harbor. The
observed periods shorter than 3 hours could be considered in terms of the
possible Helmholtz oscillation modes given in Table 2 since all of the above
9
-------
periods at Allouez and Drills appear correlated over times on the order of
one day. The observed 2.3 hour harbor oscillation corresponds, perhaps
fortuitously, with the period produced by three coupled oscillators (Table 2).
To examine for possible Helmholtz modes due to the action of a single
inlet, the results of Seeling and Sorensen (1977), who propose a i-Ielmholtz
period of 1.4 hours for the north and the south inlet of the Duluth harbor,
were considered. The observed period of 1.5 hours at Allouez could corres-
pond to this value. However, Seeling and Sorensen took the reservoir area
for the oscillator as an area somewhat smaller than the outer harbor area
which may not be justified. The above authors define a frictionless inlet—
bay Helmholtz period as:
T = 2 [ (L+L) y/(gA )]½
where T is the frictionless inlet—bay Helmholtz period and
L = _(B/ ) n [ rB/(gdT2)½]
where d is the height of water above equilibrium at the entry, and B, L, and
Ac are respectively the width, length, and cross—sectional area of the inlet.
This equation for T reduces to the single oscillator solution given by w in
equation (1) if L is neglected. L is small for the Duluth—Superior Harbor.
Using equation (1), the calculated period for a single inlet oscillator
is 1.7 and 1.9 hours for the north and the south inlets respectively, if the
effective bay reservoir area is taken as the area of the outer harbor alone.
However, it is more 1.ikely that the harbor inlets act as an oscillator
together. The coupled mode for the two inlets is 2.1 hours. It will be seen
later that this period is much closer to an observed long term 2 hour oscil-
lation predominating in the harbor. On the other hand if the two lake inlets
acting as coupled oscillators are considered with Abay equal to the outer
harbor area alone rather than the entire harbor, equation (1) yields a period
of 1.3 hours, which is reasonably close to the 1.5 hour period observed in
spectral analysis of a daily water record for the outer harbor. This period
appears to damp out farther into the harbor.
To determine if indeed the Helmholtz oscillations have a dominant effect
on mass movement in the harbor or merely produce short term distortion of
lake level oscillations, long term water level fluctuations in the harbor
were examined and compared to the seiche periods of Lake Superior. The
spectral analysis on a continuous water level record of 5 weeks, a time
comparable to flushing time for the harbor (Section 11), would bring out the
persistent frequencies while the short term resonances would average out to
relatively low values. The results from the analysis of long term water
level records are shown in Figure 2. The long duration periods in Figure 2
show highly correlated peaks at 2.0, 3.8, 4.8, and the fundamental 7.9 hour
10
-------
16.
12
-8
-4.
0 _
Outer harbor
- — Inner harbor
0
1
><
a.
C
0
0
m
a.
U
0
0.
U,
Figure 2. Fourier analysis of water level data showing the long term
oscillation periods in the harbor and Lake Superior.
I I I I I
1O 12
‘I
I
I
/
I I
2 4 6 8
Hours
-------
frequency. These periods correspond closely to the lake level oscillation
modes found by Mortimer and Fee (1976). It therefore appears that the
persistent oscillations in the harbor and therefore the long term transports
are largely due to lake level fluctuations, with persistent sympathetic ring-
ing apparent at the two hour period, which does not damp out sufficiently
fast and contributes appreciably to the flushing exchange for the harbor.
The relative strength of the oscillation peaks shown in Figure 2 depends
on the action of the harbor as a filter and the selective excitation of
Helmholtz frequencies by the lake level fluctuations of similar periods. To
examine which of the calculated Helmholtz frequencies are allowed, the fact
that the water levels at the lake inlets oscillate in—phase (to within 5
minute resolution) can be used as a condition on the excitability. Such
calculations are outside of the interest here, however, it is pointed out
that the eigen vectors for the Helmholtz frequencies with periods larger than
one hour all satisfy the in—phase condition at the lake entries.
12
-------
SECTION 5
HYDRODYNAMIC MODEL
A numerical model of transports provides a useful tool in assessing
relative changes in dispersion of pollutants under a variety of seiche condi-
tions. Here a hydrodynamic model is developed and verified. This model will
subsequently serve as the basis for development of a water quality model and
discussion of dispersal of contaminants.
For this project transports for representative seiche conditions were
calculated. Since there are no pronounced effects on transports due to
factors other than river through—flow and the lake level oscillations,
representative conditions for the harbor could be simulated by running of the
model for a few specific data periods when water level fluctuations were
characteristic of the usual ranges of seiche amplitudes. The actual water
levels at the lake inlets and the St. Louis River discharge rate served as
the boundary conditions for the hydrodynamic model. The numerical grid for
the model consisted of junctions connected by straight channels, which aver-
aged about 600 meters in length, Figure 3. The channels covered the dredged
and shallow areas of the entire harbor from Lake Superior to Fond du Lac.
In a sense the grid, consisting of one—dimensional channels, provided a
pseudo two—dimensional model, where interaction between channels was allowed
at the junctions. The channels and junctions were numbered consecutively
following the scheme of McElroy and Chiu (1974). The hydrodynamic model was
based on the Columbia River Model proposed by Callaway, Bryan, and Dittsworth
(1970). The basic equations are the equations of motion (2) and continuity
(3) given by
(2)
+- 0, q = Au (3)
where x = distance along the longitudinal axis of the river
t = time
u(x,t) = velocity in x direction
13
-------
Figure 3. Hydrodynamic model grid for the inner harbor.
I-a
-------
h(x,t) = water stage
g = gravitational constant
k = frictional constant
b = width of river
q river flow
A = cross sectional area of river
Equations (2) and (3) are written in a finite difference form, the river is
schematized, i.e. depths, areas (A), widths (b), roughness coefficients (k)
are determined, initial conditions are specified and u and h are solved for
by the “leapfrog’ t method. In the leapfrog method, the initial conditions of
velocity and stage are read into the computer along with boundary conditions.
Velocity and flow are computed from the momentum equation. The computed flow
is substituted into the continuity equation to obtain a new stage elevation
which is then used in place of the initial condition to obtain new velocity
and flow values. The new flow obtained is again substituted into the contin-
uity equation and the process “leapfrogs” until the cycle is complete.
When the equations are applied to the network, the following assumptions
are made:
1. Acceleration and momentum transfer normal to the flow
direction is negligible.
2. Wavelength is at least twice the channel depth.
3. Coriolis and wind forces are negligible.
4. Each channel is straight and has a constant cross
sectional area and depth.
In a finite difference form, equation (2) becomes:
LW H.
i ii Vl 1
— --V. L i. -K V 1 V _g---j—
1 i
where v = 1 th channel velocity
At = time step
15
-------
= velocity gradient in channel i
B = channel width
1
A = channel cross sectional area
1
L. = channel length
H. = difference in stage between ends of channel
1
K. = frictional resistance coefficient
1
2
— gn
R 413
n = Manning roughness coefficient
R = hydraulic radius
Equation (3) becomes
H. Q.
= __ .j_
t A
J
where = head of junction j
J
Q. net flow into j during a time step
A. = junction surface area (constant)
Equations (4) and (5) are then solved using a two step Runge—Kutta integra-
tion method. Manning’s coefficient, n, was adjusted to give the best
agreement between measured and calculated values of velocity and stage. The
value of n ranged between 0.020 and 0.080, with the larger values found in
the area east of the Blatnik Bridge, where the channel structure was most
complex and flow rates in the shallower parts of the system were more
important because of the harbor configuration and the relative abundance of
the shallow areas. The time step was chosen to satisfy the Courant Condition
for numerical stability. Applying this condition yielded a step of about 55
seconds. When the time step in the model was varied from 10 to 60 seconds
there was no significant difference in the results for transports. For steps
larger than60 seconds the model became unstable.
A comparison of calculated and measured water currents and water levels
from 11AUG76 are shown in Figures 4 and 5 for three of the locations marked
16
-------
C)
C)
E
0
0)
C)
0.
U)
C .)
a,
E
C.)
a)
a)
0.
(I)
Figure 4.
Calculated and measured currents up and
down river from the ORTRAN coal dock.
17
Time (hours)
-------
E
a
Ci
I
5
E
U
0
I
-5
-10
0 2 4 o 3 10 12 14 1 16 2
Time hours)
Water Levels at Drills
-- - Measured
10 — Calculatet
Time (hours
Figure 5. Comparison of measured and
calculated water levels.
VJz ler L ets Alk ue2
10
Harbor Water Levels at H&Iet
- - - Measured
— Calculated
18
-------
in Figure 1. As expected the measured and calculated values for water level
follow each other closely at Allouez Bay which is located near Lake Superior
and is the water level reference point. The values depart somewhat as one
moves into the harbor. The calculated and measured currents and water levels
have the same phase and are generally within 20% of each other in magnitude.
Table 3 shows some typical flow rates for various channels of interest. It
TABLE 3. CALCULATED CURRENTS FOR VARIOUS CHANNELS OF INTEREST
Absolute Minimum, Average, and Maximum Speeds (cm/sec)
3.0 cm Seiche 15.0 cm Seiche
Location Mm Av Max Mm Ày Max
Superior Entry * —7.9 2.0 11.5 —30.4 1.9 34.6
Duluth Entry * —16.6 1.9 19.4 —41.9 2.5 45.8
Superior Front Channel * —3.0 1.1 5.0 —11.8 1.2 14.1
Duluth Harbor Basin —2.1 0.4 2.7 —5,3 0.5 6.1
Blatnik Bridge * —17.1 3.4 22.3 —45.4 3.7 50.0
Sewage Plant —15.6 3.2 20.5 —42.0 3.4 45.7
Coal Dock —7.7 1.6 10.0 —20.6 1.6 22.1
Cross Channel —0.9 —0.0 0.8 —1.9 0.0 1.8
North Channel * —9.0 1.9 11.8 —24.4 2.0 25.7
South Channel —5.1 2.1 8.8 —19.0 1.8 20.8
Arrowhead Bridge * —8.4 2.7 12.6 —26.5 2.5 28.2
Drills* —6.4 3.0 11.3 —23.2 2.6 24.7
Oliver Bridge —1.5 2.7 6.5 —8.9 2.7 11.9
Fond du Lac * 14.3 15.6 16.7 11.9 15.5 18.5
* Insitu current measurements made In channel.
can be seen that for high seiche conditions currents exceeding 20 cm/sec
occur in many channels. Currents in excess of 15 cm/sec are of interest in
19
-------
consideration of resuspension and transports of suspended solids (Sundborg
1956).
The numerical model predicts currents in all channels as a function of
Lake Superior seiche height and St. Louis River flow rate. This information
is necessary in the study and simulation of the dispersion of particulates
from the coal dock.
20
-------
SECTION 6
WATER QUALITY MODEL
In constructing the model for transport of suspended solids, each
channel in the hydraulic model is subdivided longitudinally into 20 equal
sections in which complete mixing is assumed during each integration time
step of the sediment transport model. This subdivision of the channels is
necessary to obtain realistic approximation to actual mixing processes
(Csanady 1973, Galloway and Vakil 1977).
In the water quality model the results from the hydrodynamic model are
used to predict the spatial distribution of a substance with time. The
general differential equation which describes the mass transport is:
2
D - — - -- ÷ E S
t L, 2 x
dx
where C = concentration of the substance
U = velocity of the stream flow
DL = longitudinal dispersion coefficient
t = time
x = distance along the longitudinal axis of the stream
S sources or sinks of the substance being modeled
The first term on the right represents the mass transport due to longitudinal
dispersion or diffusion, and the second term represents the transport due to
advection. For a conservative substance, there are no sources or sinks other
than inflow or diversions. Since there are no diversions, the transport of a
conservative substance can be represented as:
- =_U - +DL- -4+input (7)
21
-------
The mass transported into junction j from junction j+l through channel ± due
to each of these three processes, advection, diffusion, and input, can be
represented in finite difference form by equations (8), (10), and (11).
AM = — U A,AtC (8)
a ii a
where t M = mass transported by advection
U. = speed in channel i
A. = cross sectional area of channel I
1
C concentration of advected substance, chosen to be the
a concentration in junction j+1, which is the upstream
concentration.
= — DLAit t(C.+l C.)/xi
where Md = mass transported by diffusion
C. = concentration in junction j
= length of channel -
Using an expression for the dispersion coefficient derived by Orlob (1959)
the equation becomes
= K Q 1 IR. t (c + 1 _C )/x± (10)
where K = diffusion constant
= flow in channel I
R. = hydraulic radius of channel i
The absolute value of the flow is used in order to Insure that the mass will
be diffused from the junction with the higher concentration to the junction
with the lower concentration. From several test runs it was determined that
the advective transport Is the dominant transport process. A value of 10.0
was chosen for K, which gives average values of DL that are comparable to
those found by Orlob (1959). Finally,
AM = C. QIN. t (11)
t in 3
22
-------
where Mt = mass transport through inf low
C. = concentration of inflow
in
QIN, = flow rate of inflow to junction j
Thus for each time interval i t, the mass change for any junction j can be
written
= E( M + t M + M ).
j . a d ti
i
= [ UiAiCa t+C Qi Ri(Cj+i_Cj)/Xi] (12)
+ C. QIN.L t
in 3
where the sum over i indicates the sum over all the channels
connected to junction j.
The concentration of a substance in each junction can then be found by
dividing the mass in that junction by the junction volume, which is defined
as the sum of the volumes of the channels connected to it.
The solution technique for the water quality model is as follows.
1. Initial and boundary conditions are specified for the system.
2. Average hydraulic parameters, which were obtained from the
hydrodynamic model, are specified. These parameters include
channel speed and flow as well as junction head.
3. Mass transfers due to advection and diffusion are made
between all the junctions.
4. If the substance is not conservative the mass in each
junction is decayed using an appropriate decay coefficient.
The only decay process that was considered is settling.
Resuspension of bottom sediments is also accounted for at
this time.
5. Mass due to inflow is added to each junction where it occurs.
6. The volume of each junction is adjusted according to the
surface elevation or head change from the hydraulic input
data.
23
-------
7. The concentration in each junction is found by dividing
the total mass by the new volume of the junction.
8. Results are stored for future analysis.
9. Steps 2 through 8 are repeated for each time step until
the model time period is completed.
The time step At in the water quality model was chosen to meet the
following conditions.
1. At is an integral multiple of the time step for the
hydraulic model, which was 60 seconds, so that the hydraulic
input data for each quality time step would be an average
of an integral number of hydraulic cycles.
2. At is an integral division of the 8 hour seiche mode,
which is the dominant driving force in the model.
3. At is the largest time period which would not allow water
to completely travel the length of any channel during a
single time step.
The values of At for each seiche amplitude modeled are shown in Table 4.
TABLE 4. TIME STEPS FOR VARIOUS SEICHE AMPLITUDES
IJSED IN WATER QUALITY MODEL
Seiche Amplitude Time Step
(cm) (see)
0.0 3600
3.0 1800
6.0 1200
15.0 900
Boundary conditions which had to be determined were: the concentrations
of each modeled substance in Lake Superior at both the Duluth and Superior
entries, and the concentration at U.S. Highway 23 in Fond du Lac, the
24
-------
upstream end of the model grid. An initial distribution of concentrations
throughout the harbor also had to be determined for each modeled parameter.
This data was obtained from sampling cruises.
To ensure that the water quality model was realistic, a conservative
parameter was modeled in time. Specific conductance was ideal for this
purpose because it is a quickly measureable conservative parameter. The
specific conductance is normally a factor of two higher in the harbor than
it is in Lake Superior. Thus, influx of lake water into the harbor can be
easily detected in the conductance measurements. Furthermore, since the
conductance has substantial variations over the harbor due to known input
sources, the parameter serves well for verification of the water quality
model in time and for verification of the simulation of dispersion processes
from a point source.
Continuous measurement of specific conductance near the Blatnik Bridge
is performed routinely by the Western Lake Superior Sanitary District
(WLSSD). As a test of the water quality model, conductance was modeled at
the Blatnik Bridge over a fifteen day period from 03JUN77 to 18JUN77. The
results from the model were compared with the WLSSD conductivity data.
The boundary value and the initial condition for conductance distribu-
tion were derived from data obtained during shipboard sampling of the entire
system. Measurements were taken at three depths and the actual conductivity
distribution for the model was obtained from the depth averaged distribution
of the in situ measurements. Sewage treatment plant discharges are the major
sources of conductivity in the harbor. To determine the magnitude of these
sources, the discharge rates of the sewage treatment plants in the harbor
were obtained from WLSSD records. Figure 6 shows the daily variation in the
discharge rate at the main treatment plant located near the Blatnik Bridge.
The daily flow rates of the St. Louis River for the model period were
obtained from Minnesota Power and Light Company data.
A comparison of the actual and simulated conductivity near the Blatnik
Bridge is shown in Figure 7. The modeled data was plotted for a 6 hour
average, short enough to resolve the fluctuation in the Duluth sewage plant
output. The transport time from the sewage plant to the monitoring station
is on the order of one day, thus the fluctuation in specific conductance at
the Blatnik Bridge due to the sewage plant output was small because of
dispersion. Good agreement was obtained between the measured and simulated
conductances with an average error less than 8%. Precipitation during the
model period, Table 5, accounts for some of the wider departures between the
modeled and measured values.
The water quality model was constructed primarily for examination of
the dispersion of contaminants arising from the operation of the ORTRAN coal
facility. For further model verification, simulations were made of the
Duluth sewage plant output for parameters such as dissolved solids and
phosphates. Chemical analysis (performed by others associated with this
program) attempted to identify a leachate which would be modeled so as to
study the effects of the coal facility on the environment. It appeared,
however, that leachates were a minor environmental problem. As a result,
25
-------
discharge distribution Duluth
8
av. disch rate 074 m ’s
0
— 6
0 )
U)
4.
C’
a)
0
a)’
0 2 4 6 S 10 J2 14 16 18 20 2 2 24
hours
Figure 6. Daily variation in Duluth Sewage Plant discharge rate.
-------
Figure 7. Calculated and measured specific conductance at west Blatnik Bridge junction.
I ’ . )
days
-------
TABLE 5. PRECIPITATION RECORDED DURING WATER QUALITY SIMULATION PERIOD
Model Airport Precipitation
Day (cm)
1 1.12
2 0.0
3 0.51
4 Trace
5 1.75
6 0.13
7 0.0
8 0.30
9 0.0
10 0.0
11 0.0
12 Trace
13 Trace
14 0.38
15 0.28
the dispersal of a hypothetical conservative leachate from the coal dock was
modeled to determine contaminant residence time and the dispersion character-
istics for the harbor. These results will be discussed later.
With apparent absence of a leachate problem, attention was turned to
the investigation of the direct contamination of harbor by coal particulates.
It was evident that fine particulates are a contaminant which is directly
put into the harbor, at least at ship loading. The dispersal of particulates
could be modeled without detailed apriori knowledge of their chemical
behavior or environmental impact. The spillage at ship loading could be
treated as a point source of particles. However, the windblown material from
the pile constituted a broad source whose distribution would have to be
28
-------
determined as a boundary condition for modeling purposes. At the outset of
the research program it was generally believed that windblown dust from the
coal pile would be negligible because of dust abatement practices which
would be undertaken in the operation of the coal storage facility. However,
soon after the facility came into full operation it was evident that the
windblown dust from the coal pile was a significant source of particulate
contaminant. It was also learned from the machate studies part of the
research program that the fine particulate appeared actually to be the most
significant form of coal related pollutant to fish (Carison and Caple 1978).
Thus, the choice to focus on the dispersion of particulates as an obvious
source of coal related contaminant appeared well justified. However, before
dispersion of coal particulates in the harbor could be simulated, the nature
of the sources and the settling rates had to be examined.
29
-------
SECTION 7
WINDBLOWN COAL DUST
In the process of acquiring coal particulate data, a number of interest-
ing observations were noted concerning conditions which effect the windblown
dust source. One of the most surprising was the fact that downwind concen-
trations generated by the dumping of coal from the overhead conveyor to the
coal pile were usually less significant than the amount of coal particulates
resulting from grooming the coal pile with caterpillar tractors. Particu-
larly noticeable were the large dust plumes generated by either the cater-
pillar tractors or the earth—movers maneuvering near the base of the pile.
Since two caterpillars were at work nearly every time the coal pile was
observed, the caterpillar operatioft appeared to be the most consistant source
of particulates. When the caterpillar tractors ceased operation, the particu-
late level would usually fall below one—fourth its previous value but would
not completely drop to the background particulate level upwind from the
facility. Thus the coal pile appeared to put out particulates at low levels
on almost a continuous basis. Besides the grooming of the pile, sudden
changes in wind direction also gave rise to visually dense dust plumes.
After a rain the dust count would stay quite low for one to two days.
However, specific measurements on the behavior of the source as a function of
climatological factors such as isolation and humidity were not made. Contin-
uous monitoring and surveillance is necessary to describe the source
accurately on an annual basis. The source as a function of winds only was
examined in order to estimate roughly the annual magnitude and distribution
of the source.
In considering particulate dispersion as a function of winds, first the
concentrations and the deposition of particulates f or a specific wind event
was considered. The northeasterly wind was chosen for the purpose of
detailed study because such wind produced coal dust plumes over a flat empty
terrain where extensive measurements could be made and where access to good
sampling sites was not limited by private property restrictions. Further-
more northeast winds carried the windblown material over environmentally
significant parts of the harbor where recreational areas and fish spawning
areas are located and where the windblown particulates would constitute a
major direct input of coal related pollutants into the harbor.
To establish the magnitude of the windblown dust source, measurements
were made on the particle concentrations in the air, particle size distribu-
tions in air and water, and the rate at which particles deposited in collec-
tion buckets, on slides, and in water—filled cuvettes. The contents of catch
30
-------
buckets were analyzed for total particulate and chemical composition.
Taconite dust and grain dust for samples near the grain elevator were the
major background material. The slides were examined microscopically for
particle size distribution, while the cuvettes were analyzed optically for
particle size, concentration, and identity based on light scattering (Diehl
1978). The collection buckets provided a measure of the long term accumula-
tion of coal dust. The slide and cuvette sampling was performed during
visible dust plumes in conjunction with measurement of winds. The cuvettes
were used in relatively short duration measurements where the distribution
of sampling points was adjusted in the field to measure quickly the distribu-
tion of coal dust concentration in air and its fallout on a water surface.
To estimate the total input of particulates into the harbor it was
necessary to extend the measurements made at the sampling stations to the
entire area. For this purpose the dust plume was modeled numerically.
The basis for most mathematical treatments of dust dispersion was the
familiar steady state diffusion equation given by
K +- -- K
x y y y z z
where C is the concentration, u is the wind speed, and K , K are the hori-
zontal and vertical eddy diffusivities respectively. The coordinate system
is chosen so that the x-axis extends horizontally in the downwind direction
and the z—axis extends vertically from the earth’s surface. Both the source
and the horizontal wind field are assumed to be time independent and diffu-
sion in the direction of the wind is ignored.
One of the most widely used solutions of this equation is the Gaussian
plume formulation for a point source, commonly expressed as
C = 2r ao [ exp [ _½( )2]] [ exp [ _½( )2I + exp [ -½( ) (13)
where the diffusion parameters are usually written as
b d
o = ax and a = cx
z y
and where Q is the source intensity and H is the height of the source above
the earth’s surface defined at z = 0. The solution assumes a spatially
independent wind speed and the eddy diffusivities are only allowed to vary as
powers of the downwind distance. For a single point source, the diffusion
in the crosswind and vertical directions produces a Gaussian concentration
profile about the plume centerline. For sources which cannot be
31
-------
realistically represented by a point source, multiple point sources can he
employed and their contributions summed to yield a close approximation to the
actual source configuration.
One of the major advantages of air quality models based on this solution
is that. only a few input parameters are required. Furthermore, since a large
number of investigators have used such models, the diffusion parameters,
and o , have been studied and empirically determined under a variety of
conditions including the various classes of atmospheric stability (Pasquill
1962, McElroy 1968, Koch 1971, 1973, 1976, Turner 1970, and Busse and
Zimmerman 1973).
In its most widely used form the Gaussian plume model assumes total
reflection of the plume at the earth’s surface. In other words, the loss of
pollutants due to deposition or reaction at the surface is neglected. Since
this assumption is difficult to justify for particulate dispersion, the
diffusion equation was solved again using for a boundary condition that the
deposition at the surface is proportional to the concentration at the
surface,
Surface
= —K — = —K aC(z=o)
Flux z z z
z=o
where a is the constant of proportionality. For the case of constant eddy
diffusivities, K and K , the correct solution can thus be shown to he
y z
exp [ — 1 ’-— —]
2rx/KK 4x (14)
u(z—H) 2 u(z+H) 2 I aX+(z+H+X) 2 ,
exp [ — 4Kx i + exp [ — 4Kx — 2 c j exp [ - 4Kx jd
Although the equation is now further complicated by the deposition term, the
additional integral can be readily evaluated by computer using a series
approximation. When the source is located on the ground, 1-i = 0, this solu-
tion reduces to that given by Smith (1962) for one—dimensional case along the
z axis. Good fits to observed particulate deposition rates were obtained
using this solution.
Figure 8 shows a contour plot of predicted deposition rates downwind
from the coal pile for a 6.8 rn/sec (15 mph) northeast wind on 17APR78. The
model was calibrated with in—the—field measurements of particulate concen-
trations in air including the upwind observations used to estimate and
extract the background concentration. Dry—fall sampler measurements analyzed
microscopically were used to correlate the air dust concentration in air with
the mass deposition rates. The rates predicted by the model were then scaled
to yield the observed mass deposition rates. The results of the model
32
-------
Figure 8. DistributiOn of coal dust fallout after 1 day of 6.8 in/s northeasterly winds.
-------
yielded distribution of dust fallout and the deposition constant c quite
close to the observed dust deposition rates at sampling stations.
It should be pointed out, however, that dust deposition rates over
harbor water may differ from the deposition over flat field where most of the
calibration measurements were taken. For instance, placement of collectors
at various heights and near obstacles appeared to produce significant varia—
tions. Furthermore, the solution to the diffusion equation which was used
to model coal particulate dispersion is based on a few questionable assump-
tions. For example, in order to extract a solution, both the wind speed and
vertical eddy diffusivity K were held constant, yet both are actually func—
tions of altitude. However, as pointed out by Smith (1962), little error is
introduced by using a constant wind velocity except perhaps at short range.
The vertical eddy diffusivity Kz is difficult to represent mathematically
and no general solution of the diffusion equation has yet been found..
Depending on weather conditions, K usually increases linearly from the
ground, levels off, and then decreases at large heights or near a tempera-
ture inversion (Moore 1975). Smith actually suggests that a constant
vertical eddy diffusivity K is a reasonable approximation for the case of
downwind distances greater than a few hundred meters but less than the
distance needed for the plume to reach an inversion layer. Fortunately,
most of the areas of interest around the coal facility fall within this
downwind range.
Since the model assumes a point source surrounded by uniform terrain,
various model improvements to account for the effects of turbulence due to
the coal pile itself and large nearby structures such as grain elevators are
currently being investigated. It may be necessary to change to a numerical
model employing a finite—difference scheme.
Another imperfection of the present model is that no allowance was made
for the increased deposition rate of the larger particles (> 20 p) simply
due to their increased mass. These particles are less important in terms of
the long range transport but some improvement might be also realized in the
future by breaking the particle size distribution into various categories
which could be handled independently. The results thus far, however,
provide a good estimate of the fallout of coal particulates from a dust
plume over the harbor.
By application of the wind frequency data for the Duluth harbor shown
in Figures 9 and 10, the model was employed to estimate roughly the total
yearly loading of coal particulates into the harbor. The winds were grouped
by direction every 22.5 degrees with each group further divided into three
speeds. The model was then used to predict downwind deposition for each
separate wind direction and speed with the results scaled according to the
frequency of occurrence of each wind subdivision. Finally, the individual
model results were overlayed to produce the contour map of Figure 11, show-
ing yearly deposition in the vicinity of the coal facility. A solution
where the source was taken proportional to wind velocity was also considered.
The results are shown in Figure 12 and appear to be quite similar to the
constant source solution giving the results in Figure 11.
34
-------
SEASONAL WIND DIRECTION PROBABILITY
Figure 9. Distribution of wind directions for the
Duluth—Superior harbor (4 year average).
Ln
—All 4.3% Calm
Winter 3.4% Calm
SprIng 6.7% Calm
Summer 3.4% CaIrn
Fall 3.6% Calm
C
C
U
0 .
Wind DIrectIon
-------
DULUTh WINDS SPEED DISTR UT ON
—AU
----NE
NW
‘I -----SE
- - SW
C
4O
- ‘---* ‘
L )
0’•’ /
‘ ‘I’..
10 / _
- ___ - -
Sp..d (m1 )
Figure 10. Distribution of wind speeds i or the Duluth—Superior
harbor (4 year average).
-------
Figure 11. Estimated annual distribution of coal dust fallout based on a Gaussian
plume model with constant dust source. Contour numbers represent
fallout in g/m 2 . Sampling stations are shown by letters. Accumulation
of dust in ice cover is shown by small numbers (values extended to
yearly basis).
-------
Figure 12. Estimated annual distribution ot coal dust (g/m 2 ) based on a Gaussian
plume model with dust source proportional to wind speed.
cx
-------
The source intensity for the yearly dustfall estimate was checked using
deposition rates determined from collection containers placed at six loca-
tions around the coal facility between 01MAR78 and 17MAR78. The wind
frequency data for this 17—day period was applied to the model and the fall-
out was found for these six locations. The deposition rate at each sampling
location, indicated on the contour plot of Figure 11, can be seen in Table 6
TABLE 6. COAL PARTICULATE FALLOUT (10 2 g/in 2 /day)
Collection Containers
March 1 — 17, 1978
Model predictions
Applied Winds:
March 1 — 17, 1978
Slide Samplers
Average of 9 days
in Jan. — Mar., 1978
1.1
1.2
—
3.1
3.6
5.5
2.9
3.8
6.5
11.0
6.2
15.5
3.5
6.3
10.0
56.0
48.9
92.2
along with the model results from this calibration period. The sensitivity
of the collection containers to local terrain features makes comparison
difficult. Glass—slide samplers were also deployed at these same locations
on selected days in the winter. Particles which deposited on the slides
were categorized by size and counted by use of a microscope. The results
from glass slides, which represent a total of about 9 days, are also shown in
Table 6. Although the slide deposition rates are higher than those obtained
by the collection containers, the slide results tend to be biased toward
downwind samples. Also shown in Figure 11 are the locations and results of
three ice samples. The ice samples were cut out of the harbor ice sheet in
late March. Yearly deposition rates were determined by estimating the
number of days since the ice had formed and scaling the average daily mass
of coal found in the ice—snow surface accordingly. Ice sample data north
and northeast of the pile agree quite well with the model predictions. The
39
-------
value northwest of the pile is too high, due to the contribution of coal dust
from piles located at the Hallet Storage Facility across the harbor. The
influence of this source will be discussed later in the section on snow
albedo.
The model predicts that annually about 20 metric tons of coal are
carried from the ORTRAN coal facility into the harbor waters, and roughly
1.5 metric tons are deposited directly into Lake Superior. The water area
receiving the highest level of contamination is the slip located on the
northeast side of the coal pile adjoining the Burlington Northern elevators,
where nearly 3 metric tons of coal is deposited yearly. tue to both insuf-
ficient data and a few weak assumptions inherent in the model, these
estimates could be in error by a factor of two. But even the rough estimate
of windblown source shows it to be the dominant source of input of coal
particulate into the harbor and the lake. Many additional measurements need
to be made. For example, the source intensity was estimated in the model
from data taken in the late winter. The source may differ from season to
season. The changing effects of precipitation, humidity, wind, and tempera-
ture may be considerable. Operating procedures at the coal facility are also
known to be changing. In the winter, for instance, the sprinkling system
used to reduce dust from the conveyors appears to be shutdown possibly due to
freezing problems. As a result, large dust plumes are visible at times from
the overhead conveyors.
40
-------
SECTION 8
DISPERSAL OF COAL DUST FALLOUT IN TEE HARBOR WATERWAY
‘the transport of coal particulates in the harbor can be examined through
use of the water quality model. The deposition rate shown in Figure 8 corre—
sporids to a total input of 287 kg of coal particulate into the harbor per
day. By using this deposition rate as the only input source to the water
quality model, the fate of the particulates once they enter the water could
be determined if their settling rates in the waterway were known. The
settling rates for coal dust have to be approximated.
It is difficult to trace coal particulates in water and determine their
settling rates since the windblown dust is deposited over broad areas of the
harbor and the coal particulate concentrations are low in comparison to the
background suspended load. Furthermore, other coal sources contribute to the
input of coal particulates upstream from the transshipment facility. How—
ever, a realistic limit on settling rates for coal particulates can be
obtained by examining the particle size distributions and settling rates for
natural turbidity in the harbor and Lake Superior. Particle sizes for
samples collected in the field were measured optically using a forward
scattering technique. The optical method is preferable to measurements made
with electrolyte counters because forward scattering is independent of
particle shape (Kerker 1969). Figure 13 shows the size distribution for
suspended solids in the harbor, Lake Superior, material resuspended by ship
traffic, and coal dust fallout from windblown material. Coal dust fallout
and the suspended particulates in the harbor both show a break at a diameter
of 4 . A similar observation has been made by McCave (1975). It can be
seen that the coal dust particles are coarse compared to the suspended solids
in the harbor, thus, the settling rates in the harbor provide a lower limit
on the settling rate for the coal particulate.
Estimates of settling rates for in situ conditions were based on
sampling measurements and mass balance for suspended load in the harbor
channels. The distribution of suspended solids in the harbor was obtained
from rapid sampling at the stations shown in Figure 14. Samples were taken
at 50 cm below the surface, mid—depth, and 50 cm off the bottom. Sampling
stations were grouped to cover the three main sections of the harbor: Fond du
Lac to Blatnik Bridge, Blatnik Bridge to the Duluth entry, and Superior front
channel to the Superior entry. The stations within each section were sampled
at three stages of the 8 hour fundamental seiche cycle. The inner harbor
inlet at Blatnik Bridge always showed well—mixed conditions. The stations on
each side of the Inlet served as reference stations and were overlapped in
each sampling scheme. Some stratification in the suspended solids and
41
-------
V
I
0
Figure 13. Particle size distribution:
1 — Dustfall at ship loading for southerly winds,
2 — Ship traffic resuspension plume,
3 — Windblown dustfall from coal pile at 0.6 — 0.7 km and
dustfall from ship loading for northerly winds,
4 — Suspended solids in St. Louis River near runoff sources
at Pokegenia and Nemadji,
5 — St. Louis River plume in Lake Superior,
6 — Suspended solids in Lake Superior near Duluth.
Particle Diameter d (pml
42
-------
DULUTH
Figure 14. Sampling stations for water quality parameters.
The letters indicate particulate source stations.
suflhtQ’
/
C
1. s
3,0
km
43
-------
chemical parameters was evident, but generally the variation in chemical
parameters with depth indicated well—mixed conditions (U.S. E.P.A. 1976). In
the mass balance calculations, suspended solids averaged over the three
depths were used as initial values in the model. The settling rate was
varied in the model to give the best agreement between tile calculated and
measured suspended solids values. The calculated settling rate for suspended
solids in the inner harbor, Figure 15, is largely due to settling of
suspended load from the Pokegema River. Dispersion accounts for most of the
variations in suspended solids well away from the particulate sources.
Nemadji River input near the Superior entry shows about 60% load deposition
in the Superior harbor basin. It should be noted that the settling rate in
Figure 15 is not a settling rate in the usual sense since no stratification
of suspended material is allowed, although it is taken into account by the
depth averages of the sampling measurements. In the model the suspended
solids are remixed every integration step, which was chosen on the basis of
previous realistic simulations of conservative parameters.
To examine how realistically the results in Figure 15 approximate the
actual settling processes, other settling rates based on in situ measure-
ments were considered. Figure 16 shows the average removal rate of suspended
solids along the shore zone of Lake Superior. The results are based on depth
integrated values of suspended solids and turbidity measurements for several
days along transects perpendicular to an erosion bank of Lake Superior. This
erosion bank is essentially a uniform 40 km line source of red clay (Stortz
1976). After a storm, a zone of highly turbid water runs along the entire
bank. The transports are parallel to the shore, thus much of the loss of
suspended solids integrated over a cross sectional area of the shore zone
comes from settling. The uniformity of the turbid bank along the shore can
be checked from remote sensing data (Hess 1973, Sydor 1976). Comparison of
Figure 15 and Figure 16 shows that the settling rates are similar. The
settling rate for the inner harbor is slower in the first three days, that is
in the large size range. This discrepancy may be somewhat accounted for by
the fact that the inner harbor has an organic component as well as red clay
from Pokegema. Thus, the effective density for the inner harbor material is
lower for particulates exceeding the 4 p size. Over a seven—day period, the
large particles settle out, thus, the tail of the curves where they overlap
closely would be determined by the small particulates which are red clay in
both cases. The agreement at the tail of the curves in Figures 15 and 16 is
good because concentrations of low size particulates can be accurately
determined, and the well—mixed conditions are applicable. The depth and
average flow speeds in the river and the lake are also comparable, however,
it should be pointed out that the purpose of comparison here is only to
consider the general magnitude of the settling rates for in situ conditions
to determine if the calculated rate for the St. Louis River is reasonable.
It is seen that the settling rate estimated from the model is realistic.
In considering the settling rate for coal particulates, note that the
coal fallout has a relatively high fraction of large size particles. The
coarse fraction determines the mass distribution and the settling rate.
Particles exceeding 25 p contain 50 — 85 percent of the total mass. To
examine further the settling rates, Stokes velocities (McCave 1975) were
44
-------
0•
1
a)
>
0
E
0)
C l )
-o
C l)
V
0
C
a,
a
C l ,
Co
0
0
0
Ii-
Figure 15.
Calculated settling rate for St. Louis
River particulate.
Days
45
-------
SETTLING OF EROSION BANK PLUME 114 LAKE SUPERIOR
10 METER DEPTH
0.1
0.2
•
0
S
0
C
0
3
In
0
c 0.6
0
U
( .
0.7
0.8
0.9 ’-
1.0 __J __L____. I I . _____
0 1 2 3 4 5 8 7 8
D 0y5
Figure 16. Measured settling rate for red clay in
Lake Superior shore zone.
46
-------
considered. Figure 17 shows the Stokes settling rate for the Duluth harbor
particulates. The rate was obtained by assigning an average density of 1.4
(McCave 1975) to the particles in the St. Louis River, where 30% of suspended
load is organic. The resulting settling rate is a factor of two slower than
the rate previously determined from mass balance in the harbor. The discrep-
ancy arises from a variety of factors including turbulence, clumping, and
particle shapes. For coarser particles typical of coal fallout and the
resuspended material, the Stokes settling rate would well approximate actual
settling. Thus, Figures 18 and 19 are taken respectively as the settling
rates for coal particulates and resuspended material in the harbor.
Before proceeding to model particulate dispersal using the above
settling rates, it is of interest to consider for a moment the fine component
of particulate contaminants. Coal particles smaller than 5 p appear to
constitute the important fraction of the suspended material so far as toxic
effects on fish are concerned (Carison, Caple 1978). The settling rates for
the fine material can either be neglected and treated as conservative matter
for which Stokes law cannot be applied (Brun—Cottan, 1976), or the settling
rate of fine particulates can be measured in the laboratory to provide a
lower limit on the actual settling rates. The rate measured in the labora-
tory for samples of St. Louis River plume consisting largely of the fine
particulate (see Figure 13) is shown in Figure 20. It is seen that the
settling rate for particles less than 4 p is low. Thus coal particulates in
this size range would be transported to Lake Superior much like dissolved
material since this fraction constitutes less than one percent of the total
coal particulate fallout In the harbor, one would expect a model to yield a
comparable figure for the amount of coal transport from the harbor to the
lake.
Simulation of particle dispersal in the harbor waterway was obtained by
applying the settling rates given in Figure 18 to the water quality model.
Concentration of all discharges into the harbor and the particulate concen-
tration in the lake was set at zero. Initially all junction concentrations
were also set at zero. The model was then run for two days, during which
time the atmospheric deposition rate for each junction was based on Figure
8. The input for each junction was distributed uniformly over the junction
area, and was put in at a constant rate during these first two days.
The total input into the harbor was 574 kg. No settling of particulates
was allowed during the two input days since the northeasterly winds caused
a high seiche and produced enough turbulence to prevent settling. The
high seiche usually damps out after one or two days. After the two days,
the atmospheric input, Figure 21, was stopped and the model was run with
settling allowed in all channels where flow speeds were less than 15 cm/sec
(Sundborg 1956). The model was run typically for three weeks. AT the end of
that time most of the particulate either settled to the bottom or was trans-
ported to Lake Superior where it was completely removed at either the Duluth
or the Superior entry. The resulting distribution of coal sediment, due to
fallout of dust over the harbor for the northeasterly wind, Is shown in
Figure 22. If after the first two days of input, settling was allowed
in all channels at all times. a sediment distribution shown in Figure 23 was
calculated. It can be seen from Figures 22 and 23 that for high seiche
conditions an accumulation of sediment occurs in the Cross Channel in excess
47
-------
ST. LOUIS RIVER SETTLING
C
0.1
0.2
Ja3.
0.4
O.6
0
0.7
0.8
0.9
1_c . I . I
5 10 15
Day,
Figure 17. Stokes settling for St. Louis River particulate.
48
-------
S
0
E
0,
0
cn
S
C
S
0.
0 ,
c i ,
0
C
0
4 -
U
S
U.
Figure 18.
Stokes settling for windblown coal
dust from the pile and at ship loading
for northerly winds.
Days
49
-------
o SETTLING FOR RESUSPENDED MATERIAL
0.1•
-v
C
E
0.4
0
0
- v
C
C
0.
0
50.6-
C
0
0.7
0.8 -
0.9 -
1.0
o 5 10 15
Hours
Figure 19. Estimated Stokes settling for
resuspended material.
50
-------
0
E
S
0
U
U-
Figure 20. Settling in column of St. Louis River
plume particulates.
F
Days
51
-------
FALLOUT Ig/m 2 1
L .01
______ Duluth
E 1 .02
.03
.08 W st t.
_ .15
1
Cro5s
Chwnol
Figure 21. Fallout from the coal pile after 2 days of 6.8 rn/s northeasterly winds.
-------
Figure 22. Sediment distribution from 2 day fallout during northe isterly winds.
The distribution was calculated for transports due to 15 cm seiche and
no settling in high flow zones.
.01
SEDIMENT DISTRIBUTION
g/m 2 I
.03
.06
WestQats
Basin.
.09
Li ’
Cross
Channel
Arrowhad
Bridge
-------
Figure 23. Sediment distribution from 2 day fallout during northeasterly
winds. The distribution was calculated for 15 cm seiche and
SEDIMENT DISTRIBUTION
Ig/m2 1
I —1.oi
.03
1 .06
Duluth
Harbor
BasIn
Westgats
Basin
Ln
Crosa
Channel
Arrowhead
Bridge
settling in all channels.
-------
of 0.1 g/m 2 . The high seiche tends to deposit material in low flow and
shallow areas, giving the sediment distribution a patchy appearance. A
similar patchy appearance is often observed in remote sensing data for
suspended solids during the spring runoff, when red clay from the Pokegema
River appears distinct in the St. Louis River (Sydor 1978). Deposition of
coal in main channels is prevalent for low seiche conditions as shown in
Figure 24. The main channels are subsequently scoured by ship traffic. The
amount of material transported to the lake under various boundary conditions
is shown in Table 7. The maximum output to the lake is 2.6% of the total
TABLE 7. KG OF COAL PARTICULATES DEPOSITED IN LAKE SUPERIOR
FROM A 574 KG FALLOUT OVER THE HARBOR
Duluth Entry Superior Entry
3.0 cm seiche,
settling everywhere
2.2
0.02
3.0 cm seiche, settling
in low flow channels
3.2
0.12
15.0 cm seiche,
settling everywhere
12.4
0.024
15.0 cm seiche, settling
in low flow channels
15.2
6.88
load. A typical output to the lake is 0.5% of the total input, indicating
that particles less than 6 i in size constitute the bulk of the material
taken out to the lake. This fact is also reflected in the particle size
distribution for the St. Louis River plumes in Lake Superior, Figure 13.
It is evident from the results of this and the preceeding section that
the windblown source constributes a large input of particulate into the
harbor and affects areas of the harbor well upstream from the coal dock.
For instance, the area around the Arrowhead Bridge, a recreational fishing
spot, is directly subject to coal dust fallout. On the other hand, the out-
put of the coal dust from the harbor to Lake Superior is quite low as seen
from Table 7. This suggests that the harbor is a long term sink of coal
fallout. Extension of the results in Table 7 by using the yearly estimate of
fallout of coal in the harbor as the total input of particulate in the model
55
-------
Figure 24. Sediment distribution from 2 days of fallout for
northeasterly winds. Distribution calculated for
transports due to 3 cm seiche.
SEDIMENT DISTRIBUTION
Igfm 2 I
. 01
.03
U .08
.15
Duluth
Harbor
Basin
a’
Arrowhiad
Brldg.
-------
indicates that 50 — 250 kg of coal are transported to the lake annually.
This contribution of coal dust to the lake comes largely from the fraction of
fine particulate within the 12,500 kg annual input of coal dust into the
harbor during the open water season. The transport of these particles to the
lake provides input a factor of 10 lower than the direct annual coal dust
fallout over the lake of 1.5 metric tons.
57
-------
SECTION 9
PARTICULATE INPUT DURING SHIP LOADING
Each year about 100 cargoes of coal, averaging 45,000 metric tons each,
are loaded by ORTRAN into ships which dock in the channel along the north
end of the facility. During the loading process some coal is released into
the air from the loading chute and the associated conveyor, which together
constitute a point source of windblown material close to the water.
A series of measurements were made to determine the average coal input
into the harbor from ship loading. The measurements were made from a boat
which moved along the dock and the ship during the loading operation.
Samples were taken two hours before ship loading, during the ship loading,
and one hour after ship loading. Dustfall samples were collected on slides,
water—filled cuvettes, and water—filled buckets. The samples were analyzed
for particle size distribution and fallout rate. Concentrations of particles
in the air were also measured using a portable air sampler. From the average
size distribution for the particulates around the coal carrier at loading
shown in Figure 13, it is seen that the particles are coarse. The average
fallout into the harbor is estimated at 120 kg per loading. This estimate
was based on a Gaussian plume for a point source with input scaled to
correspond to the measured fallout rates. The Input varied from 20 kg —
260 kg and was highly dependent on particle size distribution and wind
direction. During a loading under southwest winds, a slick of coal 2 — 3
meters wide and 150 meters long was observed around the ship. The concen-
tration of particles in the slick after 2 hours of loading was estimated at
2 mg/cm 2 . After the loading stopped, the slick dispersed within 10 minutes,
and concentration of suspended solids near the ship dropped to within 0.5
mg/2. of the background.
To simulate the dispersion of particulates during the loading of a ship,
20 kg of fine material (with particle size distribution comparable to the
windblown material from the coal pile) was distributed over the channel
adjacent to the coal dock. The input was spread uniformly in time over 8
hours, a time comparable to the duration of ship loading. The settling rates
were taken according to Figure 18.
Figures 25 and 26 show the sediment distribution in the harbor resulting
from application of the water quality model to a single ship loading and then
extending these results to 100 ship loadings. It is seen that high seiche
and no settling conditions in the turbulent channels (Figure 25) produce
dispersal of the particulates to the adjacent channels and the entire West
Gate Basin area. The amount of material reaching Lake Superior from ship
58
-------
Figure 25. Sediment distribution resulting from yearly input of
windblown fines during ship loading — based on 15 cm
seiche transports.
-------
Dukith ii ry
i’igure 26. Sediment distribution resulting from yearly input of
windblown fines during ship loading - based on 3 cm seiche
transports.
Cosi sdiunnt f,on, 2 ton dust spUl
no ss4chs
01 t’m 2
023
045
0
L. __
-------
loading operation ranges from O.O47 of the spilled load for a low seiche to
0.26% of the load for a high seiche. A summary of the results for the trans-
port of this material is shown in Table 8.
TABLE 8. TRANSPORT OF PARTICULATES FROM THE COAL DOCK TO LAKE SUPERIOR
Percent of Particulates Entering Lake Superior
Seiche
Amplitude
(cm)
Settling in
all Channels
Settling in
Low Flow
(15 cm/sec
or less)
0.0
0.00
0.00
3.0
0.01
0.02
6.0
0.04
0.07
15.0
0.13
0.26
The concentrations of coal particles in the harbor waterway and in the
lake attributable to ship loading are low in comparison to the windblown
material. However, the source from carrier loading is important locally
since it could produce high particulate concentrations in the loading channel
over short periods of time. The input of coarse material from accidental
spill and runoff over the deck could also produce high contaminant concentra-
tion in the loading channel. However, such sources cannot be evaluated here.
In the short run, an accidental spill would be localized to the dock area.
Eventually, however, the material would be transported to the other channels
by resuspension from ship traffic.
61
-------
SECTION 10
RESUSPENSION DUE TO SHIP TRAFFIC
In order to examine the long term transport of sediment in the harbor,
especially of the coarser material, measurement of resuspension due to ship
traffic was considered and estimates of the transport of the resuspended
material to Lake Superior were made.
Measurements of resuspension due to ship traffic were made by sampling
of the water column for suspended solids and turbidity at 50 cm and 4 m
depths and 50 cm from the bottom. Sampling was performed in conjunction
with water level and current measurements. The samples were taken just
before ship passage, one minute after the ship passage, every two minutes
thereafter for 30 minutes, and every 15 minutes for the next 2.5 hours.
Turbidity above background was averaged over the cross sectional area of the
channel. The resulting settling rate determined from decrease of the
suspended solids and average turbidity in the channel is shown in Figure 27.
Extension of this curve beyond 3 hours was based on an exponential fit. The
result in Figure 27 is comparable to the settling velocity given by
Ariathurai et al. (1977) for shoal material, and by Koh andChang (1973) for
dredged material from the Great Lakes. Interpretation of the sampling
results for measurement of resuspension required measurement of water levels
and knowledge of the direction of water movement. Rapid sampling over the
shipping channel was essential.
It was difficult to estimate particle size distributions of the resus-
pended material because samples contained background material which could
not be separated from the actual resuspended material. The necessity for
dilution of samples for particle size analysis also presented problems
because coarser material settles quickly, making it difficult to estimate
accurately the percent of the total mass in the few large particles. Figure
13 shows the estimated size distribution of resuspended solids in the
St. Louis River. The corresponding settling rate calculated using Stokes
Law is shown in Figure 19. Both Figures 27 and 19 indicate settling of the
material within hours of resuspension. This result is largely due to the
coarseness of the particulates. The two settling curves provide a range of
settling rates for the resuspended material. The in situ measurements are
likely to overestimate the actual settling rate because it was difficult to
detect remnants of the well—dispersed resuspension plume obscured by fluctu-
ations in background suspended load. The Stokes settling rate based on
particle size distribution is likely to be an underestimate of the actual
settling rate because the coarser fraction was probably missed in particle
size analysis, even though precautions were taken to minimize the problem.
62
-------
V
0
E
V
I,.
0
(I )
V
C
V
0.
V
0
0
C
V
Figure 27.
Settling rate for resuspended material.
The curve was extended beyond 3 hours by
applying an exponential fit to the data.
Time (Hours)
63
-------
The water quality model was used to study the transport of the resus-
pended material. From sampling it was determined that a ship resuspends
sufficient material to produce an average concentration of 10 — 15 mg/P. of
resuspended particulates in the numerical grid channels corresponding to the
track traversed by the ship. Since the time required for a ship to travel
through the harbor to the coal dock is short compared to the oscillation
periods for the harbor, in modeling of the transport of the resuspended
material the initial suspended solids concentrations were set at 10 mg/P. in
all junctions that the ship would traverse. The concentration in all other
junctions was set at 0. The concentration of all inflow was also set at 0,
and the settling rate was taken according to Figure 27. Since the settling
rate was determined on the basis of actual concentration of suspended solids
in the harbor, the above boundary condition is realistic. Ewo dispersion
cases were tested, one where the ship entered the harbor through the Duluth
entry, and the second case where the ship entered the harbor through the
Superior entry. Generally, most ships enter through the Duluth entry, how-
ever, since the Superior entry is used occasionally, both cases were run.
Each case was simulated for two seiche levels, a low seiche of 3.0 cm
amplitude, and a high seiche 15.0 cm in amplitude. The distribution of
resuspended material after ship passage is shown in Figure 28. The resulting
sediment distribution is shown in Figure 29. The material remains largely in
the original channels where it was resuspended and usually accumulates at the
edges of the channels where it is less likely to be resuspended again. The
smaller particulates are transported outside of the resuspension channels and
accumulate In the low flow and low traffic areas.
How realistic is the simulation of resuspension? The analysis of the
harbor sediments and the associated input of contaminants into the water
column due to resuspension by ship traffic has been dealt with by Van Tassell
and Moore (1976), U.S. E.P.A. (1976), and Winslow et al. of M.P.C.A. (1976).
The M.P.C.A. report addresses itself to the actual resuspension of sediments
by ship traffic. Remote sensing data shows the presence of resuspension in
the Duluth Harbor Basin where the N.P.C.A. tests were made. The U.S. E.P.A.
found polluted sediments in parts of the harbor. In the U.S. E.P.A. investi—
gation the polluted sediments were found to be associated with fine
particulates and were found at the edges of Duluth and Superior harbor basins
where dredged channels broaden, the flows are low, and traffic is infrequent.
The contaminated sediment accumulation areas correspond well to the results
shown in Figure 29 where the new sediment areas correspond to the area of
polluted sediment identif led by the U.S. E.P.A.
An estimated kg of material is resuspended by passage of a vessel.
100 kg of the resuspended material (Table 9) flows into the lake, the
remainder settles out to be resuspended again. The fine resuspended material
entering the lake dispemes throughout the lake. The coarser fraction,
exceeding 20 ii, will be transported by the lakeshore currents to the public
beaches along Minnesota Point (Keillor 1976). Total output to the lake by
ship traffic resuspension amounts roughly to 100 metric tons of sediment per
year. This estimate was based on a traffic of 1000 large ships visiting the
harbor per year; about 10% of this traffic is due to coal carriers visiting
the transshipment facility. Presently less than 0.01% of this material is
coal particulate. The fraction of coal particulate will increase over the
64
-------
Figure 28. Suspended solids track for coal carrier entering the harbor
through Duluth entry. T is time after ship’s passage.
RESUSPENDED SOLIDS
_____ 10 mg/I T=O Duluth
__ Harbor
[ jj:: : 0.5 mg/I T=3tws. Bask
Westqat•
Basin
.11
Cross
channel
Superior
Entry
-------
Resed I mentat ion
9/rn 2
L r ] 80 Duluth
Harbor
V..: : ; 60 BasIn
L ii i 1 new sediment
West iate
Ba 1n
Cross
Channel
Arrowhead
Bridge
Superior
Entry
Figure 29. Redistribution of sediment after resuspension from a ship’s passage.
-------
years by about 2% (based on the annual input) until an equilibrium ratio
between the coal particles and other sediment in the harbor is reached.
TABLE 9. PERCENT OF TOTAL LOAD RESUSPENDED DUE TO SHIP TRAFFIC
WHICH IS CARRIED TO LAKE SUPERIOR
Output at Output at
Duluth Entry Superior Entry
Ship enters Duluth entry
3.0 cm seiche
0.11
0.00
Ship enters Duluth entry
15.0 cm seiche
0.18
0.00
Ship enters Superior entry
3.0 cm seiche
0.00
0.0003
Ship enters Superior entry
15.0 cm seiche
0.00
0.0025
67
-------
SECTION 11
POLLUTANT RESIDENCE TIME IN THE HARBOR
Examination of the dispersal of particulates indicated that extensive
areas of the harbor are affected upstream from a contaminant input source.
The contaminants appear to reside in the harbor for a prolonged time. To
determine the residence or flushing time of a pollutant in the harbor, a
simulation was made for the dispersion of 200 kg of conservative material
put in at a uniform rate along the coal dock over an 8—hour period. The
concentrations of this hypothetical substance with time was determined
through use of the water quality model. The results for two seiche ampli-
tudes are shown in Figures 30 and 31.
It is seen that the peak concentration of contaminant arrives at the
Duluth entry in 5 to 8 days and at the Superior entry in 9 to 17 days. The
peak concentration at the entries is 0.05% of the initial concentration in
the input channel. The concentration peak is spread out in time over two to
three weeks depending on the seiche amplitude. Flushing of the contaminant
into the lake is shown in Figure 32. It was assumed that once the contami-
nant reached the lake it was quickly swept away from the entry by currents.
This is a reasonable boundary condition in light of the investigation of
pollutant transports In the lake by Oman and Sydor (1978). From Figure 32
it is seen that the residence time of a conservative contaminant in the
harbor is 30 — 40 days.
It Is of interest to estimate roughly the fate of fine particulate
entering the lake from the harbor. The fine particulates may affect fish
life or the water quality at the municipal water intakes. If the laboratory
measurements of settling rates shown in Figure 20 are considered, roughly 30
percent of the fine material injected into the inner harbor would enter the
lake. Based on results of transport models for Lake Superior it is estimated
that the particles could subsequently end up at the Duluth and the Cloquet
water intakes, where their concentration would range from lO to 10—10 of
the average concentration at the input channel (Oman and Sydor 1978).
68
-------
Figure 30.
Contaminant concentration at lake entries for a
200 kg conservative input at coal dock (15 cm
seiche conditions).
69
Mod•I Tim. IDaysi
-------
Figure 31.
Contaminant concentration at lake entries for a
200 kg conservative input at coal dock (3 cm
seiche conditions).
70
Model Time IDaysi
-------
C
x
w
C
Figure 32.
Accumulated output at lake entries of a
conservative contaminant initially put in at
the coal dock.
Mod•I Tims Daysi
71
-------
SECTION 12
SNOW ALBEDO
Dustfall quickly deteriorates the reflectivity of snow cover. One of
the important consequences of the dispersal of coal dust is its effect on the
energy budget for the environment. On a local scale the relationship between
snow albedo and the heat balance in the Duluth—Superior harbor is discussed
by Sydor (1978). In view of the pending world—wide conversion to coal as the
major source of energy, change in albedo also has an implication on the heat
balance on a global scale (Physics Today, News , October 1977). Thus the
effects of coal dust on the environment has a long range environmental impor—
tance.
The relationship between snow reflectivity and dustfall was investigated
here as a method for determining the distribution of dustfall as a function
of winds. Albedo monitored through remote sensing also provided a much
needed method for surveillance of fugitive dust sources since private
property restrictions generally made location of sampling stations difficult.
Measurements of dustfall distribution were especially needed for validation
of dust dispersal models which were used in obtaining the estimates for the
total magnitude of the dust sources.
In considering albedo as a measure of accumulated dustfall, it was
necessary to perform detailed measurements regarding the effects on reflec-
tivity due to temperature, insolation, and the snow aging produced by the
global dust background. So far, only preliminary results are available here.
A series of measurements on snow reflectivity were made using Science
Associates Model 615 solarimeters, one facing up and one down. The radiom-
eters were calibrated against an Eppley 8-48 pyranometer. In determining the
relationship between snow reflectivity and the particle deposition rates,
knowledge of particle size distributions was also needed to determine the
effective cross sectional area of the accumulated fallout. The total cross
section was correlated directly with snow reflectivity. Measurements for
particle sizes and deposition rates were performed at several stations along
a southwesterly transect in an open field downwind from the coal pile. The
relationship between the cross sectional area and mass deposition is shown in
Figure 33. The relationship between the albedo and the cross sectional area
is shown in Figure 34. If a constant particle deposition rate is assumed,
the results in Figure 34 are comparable to those given by Dirinhirn (1975).
Figure 34 can also be interpreted in terms of pigment scattering theory
(Kortum 1969).
72
-------
S
0
C)
U
0
C)
Figure 33.
Relationship between fallout mass and cross—sectional
area for coal particulates deposited over snow.
-4
Coal Mess jjig/cm 2 I
-------
Figure 34. Relationship between snow albedo and percent of snow surface
covered with coal.
0
0
V
-------
Measurements of albedo were made at some of the stations used in dust
plume modeling. Table 10 shows a reasonable comparison between measured coal
TABLE 10. COAL DUST DEPOSITION (pg/cm 2 /day)
Station
Slides
Albedo
Downwind
Background
Downwind
Background
1
8.0
3.6
8.1
4.2
2
10.0
3.7
8.6
4.5
3
24.0
3.1
19.0
4.8
4
116.0
8.1
54.0
7.4
deposition rates and the deposition rates calculated from the albedo measure-
ments and the relationships in Figures 33 and 34.
To examine the application of albedo measurements for determination of
dustfall distributions, Landsat satellite data was considered. The satellite
readings were calibrated using Rice Lake as a reference target. The lake,
located 20 km northwest of the contamination sources, provided an excellent
calibration site because of the uniformity of its spatial and temporal
distribution of albedo. It was found that there is a sharp natural decay of
albedo immediately after snowfall. This is expected on the basis of pigment
theory. There is also a dependence of albedo on temperatures (Dirmhirn
1975). However, for Duluth, temperatures are generally well below freezing.
The seasonal dependence of the reflected signal at Nadir for the 81% snow
reflectance at Rice Lake is shown in Figure 35. The maximum reflectance is
obtained for clear viewing conditions and fresh snow (2 — 10 days old). Rice
Lake data for clear days provides the reference reflectance for the near zero
dust deposition rate above the global dust background. The data in Figure 35
is of preliminary nature because of the limited number of available over-
flights to date.
Landsat data for 1977 provide Images clearly showing the location of
major dust sources within the Duluth harbor. For example, the low albedo
areas surrounding various coal storage and taconite storage facilities in the
harbor can be seen in Figure 36, which shows albedo determined from Landsat
75
-------
Seasonal variation in Landsat CCT tna imuin
reflectance for Rice Lake.
—I
0
0
a
(I )
‘igure 35.
-------
Figure 36. Snow albedo derived from the 09JAN77 Landsat
scene. Low albedo areas show deposition of dust
from various storage piles in Duluth harbor.
Condensation plumes show up as high albedo.
55% AIb.do
Condensation Plumes
Ii
-------
data for 09JAN77. Notice in particular the Reiss Inland Coal Storage and
Hallet Storage Facilities, where low albedo areas adjacent to the sources
outline zones where particulate fallout is high enough to reduce the albedo
to 55 percent. This corresponds to an accumulated deposition in excess of
200 ig/cm 2 of coal dust. The accumulation has occurred as a result of dust
fallout from the piles during a week of moderate northwest winds. Notice
also in Figure 36 distinct high albedo stripes. The stripes are vapor trails
from major industrial plants. The air temperature at the time of the over-
flight on January 9 was —30°C, and the winds were calm. Thus, condensation
plumes remained dense and clearly visible in the Landsat data. The vapor
plumes are not visible in Landsat data for January 10 when the temperature
was only _200 and winds increased to 7 rn/sec.
A striking case of dust deposition over the harbor ice cover is seen in
the 27JAN77 Landsat image, Figure 37. This deposition resulted from 10 — 15
rn/sec northwest winds on 26JAN77. The image indicates that coal piles on the
Duluth side of the harbor may indeed be the largest contributors of coal dust
to the harbor waterway and Lake Superior.
The images show the potential for the use of remote sensing data in
studies of airborne fallout. For instance, it can be seen from Landsat
images that the fallout from taconite and coal piles in Duluth extends over
the ice across the harbor to Superior, Wisconsin, showing that numerous
sources within the harbor may affect the measurements at sampling stations.
Thus, care had to be exercised regarding winds when sampling was performed.
Harbor ice cover and its breakup constitute the most important source of
coal particulate contaminant into Lake Superior. Results of Section 7
indicate that yearly fallout over the ice from ORTRAN is on the order of
4 — 5 tons. The fallout is trapped in the ice cover and is transported to
the lake at spring breakup. This is by far the largest input of coal dust
into Lake Superior from the ORTRAN transshipment facility.
78
-------
Figure 37. Snow albedo derived from the 27JAN77 Landsat
scene showing dust deposition from Duluth
storage piles under action of 10 — 15 ni/s
northwest winds on 26JAN77.
79
EIJ 56%
69% AIb do
- :_ : :
y-Th-
-------
REFERENCES
Ariathurai, R., R. C. MacArthur, and R. B. Krone. 1977. Mathematical
Model of Estuarial Sediment Transport . Technical Report D—7712.
U.S.A.C.E., P.O. Box 631, Vicksburg, Mississippi 39100.
Bennett, J. P. 1974. “Concepts of Mathematical Modeling of Sediment
Yield.” Water Resources Research lO(3):485.
Bierman, Jr., V. J., and W. R. Swain. 1978. “DDT Loss Rates in Low Lipid
Fishes in the Great Lakes.” 21st Conference on Great Lakes Research,
Windsor, Ontario.
Brun—Cottan, J. C. 1976. “Stokes Settling and Dissolution Rate Model for
Marine Particles as a Function of Size Distribution.” J. Geophys. Res .
81(9) :1601.
Busse, A. D., and J. R. Zimmerman. 1973. User’s Guide for the Climatolog—
ical Dispersion Model . U.S. E.P.A. Pubi. No. R4—73—024. Prepared for
E.P.A., Research Triangle Park, N.C., December.
Callaway, R. J., and K. V. Byran, and G. R. Dittsworth. 1970. Mathematical
Model of the Columbia River . NITS, U.S. Dept. of Commerce, PB202.
Carison, R., and R. Caple. 1978. Final Report for Coal Leachate Studies.
U.S. E.P.A. Grant R803952—03—3. University of Minnesota, Duluth.
Carrier, G. F., R. P. Shaw, and M. Miyata. 1971. “Response of Narrow—
Mouthed Harbor in a Straight Coastline to Periodic Incident Wave.” J.
Eng. Mech. Division. Proc. A.S.C.E . EM6 p. 1703.
Churchill, J. H. 1976. Measurement and Computer Modeling of the Distribu-
tion of Nuclear Reactor Discharged Radionuclides in the Estuarine
Sediment Near the Main Yankee Atomic Power Plant in Wiscasset, Maine .
M.S. Thesis. University of Maine at Orono.
Csanady, G. T. 1973. Turbulent Diffusion in the Environment . D. Reide L:
Boston.
Def ant, A. 1961. Physical Oceanography . Pergamon Press:London.
Diehl, S. R., D. T. Smith, and M. Sydor. 1978. Optical Detection of Fiber
Particles in Water . Final Report U.S. E.P.A. Grant R804361.
80
-------
Dirnihirn, I., and F. D. Eaton. 1975. “Some Characteristics of the Albedo
of Snow.” J. Appl. Meteorol . 14:375—379.
Feigner, K. D., and H. S. Harris. 1970. Documentation Report — FWQA
Dynamic Estuary Model . NTIS Publ. No. PB 197 103.
Freeman, N. G., P. F. Hamblin, and I. S. Murty. 1974. “Helmholtz
Resonance in Harbours of the Great Lakes.” Proc. 17th Conference on
Great Lakes Research , p. 399—411.
Galloway, Jr., F. M., and S. J. Vakil. 1977. “Criteria for the Use of
Vertical Averaging in Great Lakes Dispersion Models.” J. Great Lakes
Res. 3(1—2):20—28.
Hess, C. S. 1973. Study of Shoreline Erosion on the Western Arm of Lake
Superior . A report to Reserve Mining Company, Silver Bay, Minnesota.
Hwang, Li—San, and E. 0. Tuck. 1970. “On the Oscillations of Harbours of
Arbitrary Shape.” J. Fluid Mech . 42(3):447.
Ippen, A. T., and Y. Goday. 1963. Wave Induced Oscillations in Harbours .
Hydrolab Massachusetts Institute of Technology.
Keillor, J. P. and R. A. Ragotzkie. 1976. An Assessment of the Environ-
mental Effects of Dredged Material Disposal in Lake Superior, Vol. 1 .
Marine Studies Center, University of Wisconsin, Madison.
Kerker, M. 1969. The Scattering of Light . Academic Press.
Koch, R. C., and S. D. Thayer. 1971. Validation and Sensitivity Analysis
of the Gaussian Plume Multiple — Source Urban Diffusion Model .
Contract No. CPA 70—94. Prepared for E.P.A., Research Triangle Park,
N.C. by GROMET, Inc., Gaithensburg, Maryland.
Koch, R. C., and G. E. Fisher. 1973. Evaluation of the Mu1tiple—SOUr
Gaussian Plume Diffusion Model, Phase I Report . Contract No. 68—02—
0281. Prepared for E.P.A., Research Triangle Park, N.C., by GROMET,
Inc., Gaithensburg, Maryland.
Koch, R. C., D. J. Pelton, and P. H. Hwang. 1976. Sampled Chronological
Input Model (5dM) Applied to Air Quality Planning in Two Large
Metropolitan Areas . E.P.A. Pubi. No. 600/9—76—016. Prepared for E.P.A.,
Research Triangle Park, N.C., by GROMET, Inc., Gaithensburg, Maryland.
Koh, R. C. Y., and Y. C. Chang. 1973. Mathematical Model for Barged Ocean
Disposal of Wastes . E.P.A. 66012—73—029.
Kortum, G. 1969. Reflectance Spectroscopy. Springer—Verlag: New York.
McCave, I. N. 1975. “Vertical Flux of Particles in Ocean.” Deep Sea Res .
22:491, Pergamon Press.
81
-------
McElroy, J. L., and F. Pooler. 1968. St. Louis Dispersion Study, Vol. II —
Analjsis . AP-53, E.P.A., Research Triangle Park, N.C.
McElroy, A. D., and S. Y. Chiu. 1974. Water Quality Investigation:Duluth-.
Superior Harbor . E.P.A. Contract No. 68—01—1593. Midwest Research
Institute, Kansas City, Missouri.
Miles, J. W. 1974. “Harbor Seiching.” Annual Review of Fluid Mechanics
6:17.
Moore, D. J. 1975. “Application of CEGB Plume Rise and Dispersion Results
to Prediction Models for Ground Level Concentrations.” International
Clean Air Conference, Rotorua, New Zealand.
Mortimer, C. H., and E. J. Fee. 1976. “Surface Oscillations and Tides of
Lake Michigan and Superior.” Phil. Trans. Roy. Soc. London 281:1—61.
Oman, G. J., and M. Sydor. 1978. “Use of Remote Sensing in Determination of
Chemical Loading of Lake Superior Due to Spring Runoff.” Canadian J .
Spectroscopy 23(2) :52.
Orlob, G. T. 1959. “Eddy Diffusion in Homogenous Turbulence.” J. of
Hydraulics Division of A.S.C.E. , p. 1149.
Pasquill, F. 1962. Atmospheric Diffusion . D. Van Nostrand Co., Ltd.
London.
Santer, G. D. 1976. A Generic Survey of Air Quality Simulation Models .
University of California Lawrence Livermore Laboratory, E.R.D.A.
Contract No. W—7405—Eng.—48.
Seeling, W. N., and R. M. Sorensen. 1977. Hydraulics of Great Lakes
Inlets . Technical Paper No. 77—8, U.S. Army Corps of Engineers Coastal
Engineering Research Center, Kingman Building, Fort Belvoir, Virginia
22060, July.
Shackman, T. A. 1978. Duluth Metropolitan Area Bulk Storage Facility
Inventory and Fugitive Emission Summary . M.P.C.A. Report.
Smith, F. B. 1962. “The Problem of Deposition in Atmospheric Diffusion of
Particulate Matter.” J. Atmospheric Sciences 19:429—434.
Stortz, K. R., R. T. Clapper, and M. Sydor. 1976. “Turbidity Sources in
Lake Superior.” J. Great Lakes Res . 2:393.
Sundborg, A. 1956. “The River Kiaralven: A Study of Fluvial Processes.”
Geografiska Annaler Vol. 38 A, Sweden.
Sydor,M. 1976. Red Clay Turbidity and Its Transport in Western Lake
Superior . U.S. E.P.A. R—005l75. Final Report.
82
-------
Sydor, M. 1978. “Ice Growth in Duluth-Superior Harbor.” J. Geophys. Res .
83(C8):4074.
Sydor, N. 1978. “Analysis of Suspended Solids in Lakes Using Landsat Data.”
Canadian J. Spectroscopy 23:91.
Turner, D. B. 1970. Workbook of Atmospheric Dispersion Estimates . Public
Health Service Pubi. No. 999—AP—26.
U.S. E.P.A. Report. 1976. Report on the Degree of Pollution of Bottom
Sediments — Duluth—Superior Harbor — Sampled June 29—30, 1976 . Region
V, U.S. E.P.A.
Van Tassell, J., and J. R. Moore. 1976. An Assessment of the Environmental
Effects of Dredged Material Disposal in Lake Superior. Vol. 2 . Marine
Studies Center Publication, University of Wisconsin, Madison.
Winslow, J. 1976. M.P.C.A. Propeller Wash Study Duluth—Superior Harbor .
Minnesota Pollution Control Agency, St. Paul, Minnesota.
83
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
i2.
of Coal in the Duluth—Superior
3. RECIPIENTS ACCESSION NO.
- ._____________
5. REPORT DATE
January 1980 issuing date
6.PERFORMING ORGANIZATION CODE
Stortz
8. PERFORMING ORGANIZATION REPORT NO.
NAME AND ADDRESS
Minnesota——Duluth
55812
10. PROGRAM ELEMENT NO.
1NE625
11.CONTRACT/GRANTNO.
Grant No. R—803952
AND ADDRESS
Laboratory — Duluth, MN
Development
Protection Agency
55804
13. TYPE OF REPORT AND PERIOD COVERED
—
14. SPONSORING AGENCY CODE
EPA/600/03
from an 0RTI AN coal transshipment facility was
the input of coal dust into the Duluth harbor and to
of coal particulates to Lake Superior. A numerical
discuss dispersal of contaminants and determine the residence
the water way. The model was verified using measurements
and water quality parameters.
KEY WORDS AND DOCUMENT ANALYSIS
b. IDENTI FlEAS/OPEN ENDED TEAMS
C. COSATIFiCId/Group
55/0
62/E
67/E
63/F
64/K
80/G
Coal
Harbor
Particulates
19. SECURITY CLASS (This Reporr)
UNCLASSIFIED
21. NO. OF PAGES
94
22. PRICE
20. SECURITY CLASS (This page)
UNCLASSIFIED
EPA Form 2220—1 (R.v. 4—77)
PREVIOUS EDITION IS OBSOLETE
84
S SOVIANMENI PR NflNV OIFICI 980 —657—146/5545
------- |